id
stringlengths 9
16
| submitter
stringlengths 1
64
⌀ | authors
stringlengths 3
62.8k
| title
stringlengths 1
382
| comments
stringlengths 1
1.15k
⌀ | journal-ref
stringlengths 1
557
⌀ | doi
stringlengths 7
153
⌀ | report-no
stringlengths 1
509
⌀ | categories
stringlengths 5
125
| license
stringclasses 9
values | abstract
stringlengths 3
6.09k
| versions
listlengths 1
187
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-08-22 00:00:00
| authors_parsed
listlengths 1
3.31k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2508.12281
|
Xin Dai
|
Xin Dai, Buqiang Xu, Zhenghao Liu, Yukun Yan, Huiyuan Xie, Xiaoyuan Yi, Shuo Wang, Ge Yu
|
Legal$\Delta$: Enhancing Legal Reasoning in LLMs via Reinforcement Learning with Chain-of-Thought Guided Information Gain
| null | null | null | null |
cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
Legal Artificial Intelligence (LegalAI) has achieved notable advances in automating judicial decision-making with the support of Large Language Models (LLMs). However, existing legal LLMs still struggle to generate reliable and interpretable reasoning processes. They often default to fast-thinking behavior by producing direct answers without explicit multi-step reasoning, limiting their effectiveness in complex legal scenarios that demand rigorous justification. To address this challenge, we propose Legal$\Delta$, a reinforcement learning framework designed to enhance legal reasoning through chain-of-thought guided information gain. During training, Legal$\Delta$ employs a dual-mode input setup-comprising direct answer and reasoning-augmented modes-and maximizes the information gain between them. This encourages the model to acquire meaningful reasoning patterns rather than generating superficial or redundant explanations. Legal$\Delta$ follows a two-stage approach: (1) distilling latent reasoning capabilities from a powerful Large Reasoning Model (LRM), DeepSeek-R1, and (2) refining reasoning quality via differential comparisons, combined with a multidimensional reward mechanism that assesses both structural coherence and legal-domain specificity. Experimental results on multiple legal reasoning tasks demonstrate that Legal$\Delta$ outperforms strong baselines in both accuracy and interpretability. It consistently produces more robust and trustworthy legal judgments without relying on labeled preference data. All code and data will be released at https://github.com/NEUIR/LegalDelta.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 08:10:08 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 02:05:06 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Dai",
"Xin",
""
],
[
"Xu",
"Buqiang",
""
],
[
"Liu",
"Zhenghao",
""
],
[
"Yan",
"Yukun",
""
],
[
"Xie",
"Huiyuan",
""
],
[
"Yi",
"Xiaoyuan",
""
],
[
"Wang",
"Shuo",
""
],
[
"Yu",
"Ge",
""
]
] |
2508.12314
|
Chiranjit Mitra
|
Chiranjit Mitra
|
Synchronization Dynamics of Heterogeneous, Collaborative Multi-Agent AI Systems
|
9 pages, 6 figures
| null | null | null |
cs.MA cs.AI nlin.AO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We present a novel interdisciplinary framework that bridges synchronization theory and multi-agent AI systems by adapting the Kuramoto model to describe the collective dynamics of heterogeneous AI agents engaged in complex task execution. By representing AI agents as coupled oscillators with both phase and amplitude dynamics, our model captures essential aspects of agent specialization, influence, and communication within networked systems. We introduce an order parameter to quantify the degree of coordination and synchronization, providing insights into how coupling strength, agent diversity, and network topology impact emergent collective behavior. Furthermore, we formalize a detailed correspondence between Chain-of-Thought prompting in AI reasoning and synchronization phenomena, unifying human-like iterative problem solving with emergent group intelligence. Through extensive simulations on all-to-all and deterministic scale-free networks, we demonstrate that increased coupling promotes robust synchronization despite heterogeneous agent capabilities, reflecting realistic collaborative AI scenarios. Our physics-informed approach establishes a rigorous mathematical foundation for designing, analyzing, and optimizing scalable, adaptive, and interpretable multi-agent AI systems. This work opens pathways for principled orchestration of agentic AI and lays the groundwork for future incorporation of learning dynamics and adaptive network architectures to further enhance system resilience and efficiency.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 10:16:41 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Mitra",
"Chiranjit",
""
]
] |
2508.12316
|
Alessandro Torrielli
|
Alessandro Torrielli
|
AdS3 integrability, Sine-Gordon and fractional supersymmetry
|
16 pages, 4 figures; v2: added references
| null | null |
DMUS-MP-25/04
|
hep-th
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The massless S-matrix of the pure RR AdS3 X S3 X T4 theory is very similar but not quite the same as Fendley-Intriligator's N=2 S-matrix, in turns related to Sine-Gordon taken at a special coupling. In this short note we review the reason why supersymmetry emerges but with a fractional statistics of the particles. We then use this to obtain a very simple interpolating S-matrix between massless AdS3 and Sine-Gordon, and further find two interpolating S-matrices between these and the mixed-flux relativistic S-matrix, solving the Yang-Baxter equation all the way in-between. They are all relativistic invariant and braiding unitary, can be written in terms of particles with fractional statistics, and they might serve the purpose to embed into a parent integrable system.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 10:18:28 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 11:52:29 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Torrielli",
"Alessandro",
""
]
] |
2508.12363
|
Chenhang Xu
|
Zijing Chen, Chenhang Xu, Chendi Xie, Weichen Tang, Qiaomei Liu, Dong Wu, Qing Xu, Tao Jiang, Pengfei Zhu, Xiao Zou, Jun Li, Zhiwei Wang, Nanlin Wang, Dong Qian, Alfred Zong, and Dao Xiang
|
Structural contribution to light-induced gap suppression in Ta$_2$NiSe$_5$
| null | null | null | null |
cond-mat.mtrl-sci cond-mat.str-el
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
An excitonic insulator is a material that hosts an exotic ground state, where an energy gap opens due to spontaneous condensation of bound electron-hole pairs. Ta$_2$NiSe$_5$ is a promising candidate for this type of material, but the coexistence of a structural phase transition with the gap opening has led to a long-standing debate regarding the origin of the insulating gap. Here we employ MeV ultrafast electron diffraction to obtain quantitative insights into the atomic displacements in Ta$_2$NiSe$_5$ following photoexcitation, which has been overlooked in previous time-resolved spectroscopy studies. In conjunction with first-principles calculations using the measured atomic displacements, we find that the structural change can largely account for the photoinduced reduction in the energy gap without considering excitonic effects. Our work illustrates the importance of a quantitative reconstruction of individual atomic pathways during nonequilibrium phase transitions, paving the way for a mechanistic understanding of a diverse array of phase transitions in correlated materials where lattice dynamics can play a pivotal role.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 13:42:26 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 01:04:23 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Chen",
"Zijing",
""
],
[
"Xu",
"Chenhang",
""
],
[
"Xie",
"Chendi",
""
],
[
"Tang",
"Weichen",
""
],
[
"Liu",
"Qiaomei",
""
],
[
"Wu",
"Dong",
""
],
[
"Xu",
"Qing",
""
],
[
"Jiang",
"Tao",
""
],
[
"Zhu",
"Pengfei",
""
],
[
"Zou",
"Xiao",
""
],
[
"Li",
"Jun",
""
],
[
"Wang",
"Zhiwei",
""
],
[
"Wang",
"Nanlin",
""
],
[
"Qian",
"Dong",
""
],
[
"Zong",
"Alfred",
""
],
[
"Xiang",
"Dao",
""
]
] |
2508.12365
|
Shaowei Yao
|
Chenhe Dong, Shaowei Yao, Pengkun Jiao, Jianhui Yang, Yiming Jin, Zerui Huang, Xiaojiang Zhou, Dan Ou, Haihong Tang
|
TaoSR1: The Thinking Model for E-commerce Relevance Search
| null | null | null | null |
cs.IR cs.AI cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Query-product relevance prediction is a core task in e-commerce search. BERT-based models excel at semantic matching but lack complex reasoning capabilities. While Large Language Models (LLMs) are explored, most still use discriminative fine-tuning or distill to smaller models for deployment. We propose a framework to directly deploy LLMs for this task, addressing key challenges: Chain-of-Thought (CoT) error accumulation, discriminative hallucination, and deployment feasibility. Our framework, TaoSR1, involves three stages: (1) Supervised Fine-Tuning (SFT) with CoT to instill reasoning; (2) Offline sampling with a pass@N strategy and Direct Preference Optimization (DPO) to improve generation quality; and (3) Difficulty-based dynamic sampling with Group Relative Policy Optimization (GRPO) to mitigate discriminative hallucination. Additionally, post-CoT processing and a cumulative probability-based partitioning method enable efficient online deployment. TaoSR1 significantly outperforms baselines on offline datasets and achieves substantial gains in online side-by-side human evaluations, introducing a novel paradigm for applying CoT reasoning to relevance classification.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 13:48:48 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Dong",
"Chenhe",
""
],
[
"Yao",
"Shaowei",
""
],
[
"Jiao",
"Pengkun",
""
],
[
"Yang",
"Jianhui",
""
],
[
"Jin",
"Yiming",
""
],
[
"Huang",
"Zerui",
""
],
[
"Zhou",
"Xiaojiang",
""
],
[
"Ou",
"Dan",
""
],
[
"Tang",
"Haihong",
""
]
] |
2508.12393
|
Duzhen Zhang
|
Duzhen Zhang, Zixiao Wang, Zhong-Zhi Li, Yahan Yu, Shuncheng Jia, Jiahua Dong, Haotian Xu, Xing Wu, Yingying Zhang, Tielin Zhang, Jie Yang, Xiuying Chen, Le Song
|
MedKGent: A Large Language Model Agent Framework for Constructing Temporally Evolving Medical Knowledge Graph
| null | null | null | null |
cs.CL cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The rapid expansion of medical literature presents growing challenges for structuring and integrating domain knowledge at scale. Knowledge Graphs (KGs) offer a promising solution by enabling efficient retrieval, automated reasoning, and knowledge discovery. However, current KG construction methods often rely on supervised pipelines with limited generalizability or naively aggregate outputs from Large Language Models (LLMs), treating biomedical corpora as static and ignoring the temporal dynamics and contextual uncertainty of evolving knowledge. To address these limitations, we introduce MedKGent, a LLM agent framework for constructing temporally evolving medical KGs. Leveraging over 10 million PubMed abstracts published between 1975 and 2023, we simulate the emergence of biomedical knowledge via a fine-grained daily time series. MedKGent incrementally builds the KG in a day-by-day manner using two specialized agents powered by the Qwen2.5-32B-Instruct model. The Extractor Agent identifies knowledge triples and assigns confidence scores via sampling-based estimation, which are used to filter low-confidence extractions and inform downstream processing. The Constructor Agent incrementally integrates the retained triples into a temporally evolving graph, guided by confidence scores and timestamps to reinforce recurring knowledge and resolve conflicts. The resulting KG contains 156,275 entities and 2,971,384 relational triples. Quality assessments by two SOTA LLMs and three domain experts demonstrate an accuracy approaching 90%, with strong inter-rater agreement. To evaluate downstream utility, we conduct RAG across seven medical question answering benchmarks using five leading LLMs, consistently observing significant improvements over non-augmented baselines. Case studies further demonstrate the KG's value in literature-based drug repurposing via confidence-aware causal inference.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 15:14:03 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 05:18:31 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Zhang",
"Duzhen",
""
],
[
"Wang",
"Zixiao",
""
],
[
"Li",
"Zhong-Zhi",
""
],
[
"Yu",
"Yahan",
""
],
[
"Jia",
"Shuncheng",
""
],
[
"Dong",
"Jiahua",
""
],
[
"Xu",
"Haotian",
""
],
[
"Wu",
"Xing",
""
],
[
"Zhang",
"Yingying",
""
],
[
"Zhang",
"Tielin",
""
],
[
"Yang",
"Jie",
""
],
[
"Chen",
"Xiuying",
""
],
[
"Song",
"Le",
""
]
] |
2508.12410
|
Jun Zeng
|
Jun Zeng, Yannan Huang, Elif Keles, Halil Ertugrul Aktas, Gorkem Durak, Nikhil Kumar Tomar, Quoc-Huy Trinh, Deepak Ranjan Nayak, Ulas Bagci, Debesh Jha
|
SRMA-Mamba: Spatial Reverse Mamba Attention Network for Pathological Liver Segmentation in MRI Volumes
|
9 pages, 4 figures
| null | null | null |
cs.CV cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Liver Cirrhosis plays a critical role in the prognosis of chronic liver disease. Early detection and timely intervention are critical in significantly reducing mortality rates. However, the intricate anatomical architecture and diverse pathological changes of liver tissue complicate the accurate detection and characterization of lesions in clinical settings. Existing methods underutilize the spatial anatomical details in volumetric MRI data, thereby hindering their clinical effectiveness and explainability. To address this challenge, we introduce a novel Mamba-based network, SRMA-Mamba, designed to model the spatial relationships within the complex anatomical structures of MRI volumes. By integrating the Spatial Anatomy-Based Mamba module (SABMamba), SRMA-Mamba performs selective Mamba scans within liver cirrhotic tissues and combines anatomical information from the sagittal, coronal, and axial planes to construct a global spatial context representation, enabling efficient volumetric segmentation of pathological liver structures. Furthermore, we introduce the Spatial Reverse Attention module (SRMA), designed to progressively refine cirrhotic details in the segmentation map, utilizing both the coarse segmentation map and hierarchical encoding features. Extensive experiments demonstrate that SRMA-Mamba surpasses state-of-the-art methods, delivering exceptional performance in 3D pathological liver segmentation. Our code is available for public: https://github.com/JunZengz/SRMA-Mamba.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 15:52:54 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 06:05:00 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Zeng",
"Jun",
""
],
[
"Huang",
"Yannan",
""
],
[
"Keles",
"Elif",
""
],
[
"Aktas",
"Halil Ertugrul",
""
],
[
"Durak",
"Gorkem",
""
],
[
"Tomar",
"Nikhil Kumar",
""
],
[
"Trinh",
"Quoc-Huy",
""
],
[
"Nayak",
"Deepak Ranjan",
""
],
[
"Bagci",
"Ulas",
""
],
[
"Jha",
"Debesh",
""
]
] |
2508.12414
|
Hsiu-Chuan Hsu
|
Bo-Xin Lin, Hsiu-Chuan Hsu
|
Bulk photovoltaic effects in the Haldane model
| null | null | null | null |
cond-mat.mes-hall
|
http://creativecommons.org/licenses/by/4.0/
|
The bulk photovoltaic effect (BPVE) refers to the direct current generation in a noncentrosymmetric material under illumination and can be applied to solar energy technology. BPVE includes injection and shift currents, led by the change of velocity and displacement of wave packet during optical transitions, respectively. We derive the constraints on the conductivity tensors imposed by mirror-time ($\mathcal{MT}$) symmetry for two-dimensional systems. For the Haldane model, we show that linearly polarized light can induce shift and injection currents, which vanish under circularly polarized light as constrained by the three-fold rotation and $\mathcal{MT}$ symmetry. Additionally, due to the presence of $\mathcal{MT}$ symmetry, a separation of responses is observed in the Haldane model: one direction exhibits a time-reversal symmetry-allowed response, whereas another manifests a parity-time symmetry-allowed response. Across the topological phase transition, the injection current does not change sign, whereas shift current shows a sign flip. The vector field of the Hermitian connection in the Brillouin zone possesses vortices in the topological phase, but not in the trivial phase. Furthermore, we calculate the related quantum geometry, including Berry curvature, quantum metric and Hermitian connection, and demonstrate the microscopic quantum origin of the BPVE.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 16:01:52 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 11:01:18 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Lin",
"Bo-Xin",
""
],
[
"Hsu",
"Hsiu-Chuan",
""
]
] |
2508.12437
|
Qiang Hao
|
Qiang Hao, Shaojie Yang, Peter Yun, Jun Ruan, Shougang Zhang
|
Portable Laser-Pumped Rb Atomic Clock with Digital Circuits
| null | null |
10.1109/TIM.2025.3547469
| null |
physics.atom-ph quant-ph
|
http://creativecommons.org/licenses/by/4.0/
|
Reducing the size and complexity of high-performance timekeeping devices is an ever-growing need for various applications, such as 6G wireless technology, positioning, navigation and timing (PNT), Internet of Things (IoT), and ultrafast spectroscopy. This work presents a distributed feedback (DFB) laser-pumped Rb atomic clock, which features extraordinary frequency stability, small size and low power consumption. The DFB laser head employs a built-in isolator with a linewidth of approximately 1 MHz. For complete optical pumping of the atoms in the absorption cell, the laser beam is expanded to a diameter of 10 mm by using an optical diffuser-based beam expander. The physics package is based on a magnetron microwave cavity and surrounded by two layers of magnetic shielding. The overall volume of the optical system combined with the physics package is 250 cm$^3$. The proposed atomic clock is also designed to operate at a low temperature, whose absorption cell is maintained at 323 K. Benefiting from the lower Rb atom density, the excited atoms present a long population relaxation time of 5.8 ms. The frequency synthesizer and frequency-locked loop are implemented by digital circuits. The short-term stability of the atomic clock is measured to be $1.8\times10^{-12}\tau ^{-1/2}$ (1-100s). Our achievement paves the way for practical application of the laser-pumped Rb atomic clocks.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 17:09:30 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 03:23:11 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Hao",
"Qiang",
""
],
[
"Yang",
"Shaojie",
""
],
[
"Yun",
"Peter",
""
],
[
"Ruan",
"Jun",
""
],
[
"Zhang",
"Shougang",
""
]
] |
2508.12448
|
Yeongwoo Song
|
Yeongwoo Song, Jaeyong Bae, Dong-Kyum Kim, Hawoong Jeong
|
Uncovering Emergent Physics Representations Learned In-Context by Large Language Models
|
17 pages, 10 figures
| null | null | null |
cs.CL cs.AI cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Large language models (LLMs) exhibit impressive in-context learning (ICL) abilities, enabling them to solve wide range of tasks via textual prompts alone. As these capabilities advance, the range of applicable domains continues to expand significantly. However, identifying the precise mechanisms or internal structures within LLMs that allow successful ICL across diverse, distinct classes of tasks remains elusive. Physics-based tasks offer a promising testbed for probing this challenge. Unlike synthetic sequences such as basic arithmetic or symbolic equations, physical systems provide experimentally controllable, real-world data based on structured dynamics grounded in fundamental principles. This makes them particularly suitable for studying the emergent reasoning behaviors of LLMs in a realistic yet tractable setting. Here, we mechanistically investigate the ICL ability of LLMs, especially focusing on their ability to reason about physics. Using a dynamics forecasting task in physical systems as a proxy, we evaluate whether LLMs can learn physics in context. We first show that the performance of dynamics forecasting in context improves with longer input contexts. To uncover how such capability emerges in LLMs, we analyze the model's residual stream activations using sparse autoencoders (SAEs). Our experiments reveal that the features captured by SAEs correlate with key physical variables, such as energy. These findings demonstrate that meaningful physical concepts are encoded within LLMs during in-context learning. In sum, our work provides a novel case study that broadens our understanding of how LLMs learn in context.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 17:49:17 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Song",
"Yeongwoo",
""
],
[
"Bae",
"Jaeyong",
""
],
[
"Kim",
"Dong-Kyum",
""
],
[
"Jeong",
"Hawoong",
""
]
] |
2508.12514
|
Jaime Andr\'es Vera-Jaramillo
|
Jaime Vera-Jaramillo
|
Reconstructing Subnational Labor Indicators in Colombia: An Integrated Machine and Deep Learning Approach
|
85 pages, 23 figures, 17 tables
| null | null | null |
econ.EM
|
http://creativecommons.org/licenses/by/4.0/
|
This study proposes a unified multi-stage framework to reconstruct consistent monthly and annual labor indicators for all 33 Colombian departments from 1993 to 2025. The approach integrates temporal disaggregation, time-series splicing and interpolation, statistical learning, and institutional covariates to estimate seven key variables: employment, unemployment, labor force participation (PEA), inactivity, working-age population (PET), total population, and informality rate, including in regions without direct survey coverage. The framework enforces labor accounting identities, scales results to demographic projections, and aligns all estimates with national benchmarks to ensure internal coherence. Validation against official departmental GEIH aggregates and city-level informality data for the 23 metropolitan areas yields in-sample Mean Absolute Percentage Errors (MAPEs) below 2.3% across indicators, confirming strong predictive performance. To our knowledge, this is the first dataset to provide spatially exhaustive and temporally consistent monthly labor measures for Colombia. By incorporating both quantitative and qualitative dimensions of employment, the panel enhances the empirical foundation for analysing long-term labor market dynamics, identifying regional disparities, and designing targeted policy interventions.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 22:26:42 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 05:27:53 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Vera-Jaramillo",
"Jaime",
""
]
] |
2508.12539
|
Sandaru Wanigasekara Jayawardana
|
Sandaru Jayawardana, Sennur Ulukus, Ming Ding and Kanchana Thilakarathna
|
The Hidden Cost of Correlation: Rethinking Privacy Leakage in Local Differential Privacy
|
19 pages with 8 figures
| null | null | null |
cs.CR cs.IT math.IT
|
http://creativecommons.org/licenses/by/4.0/
|
Local differential privacy (LDP) has emerged as a promising paradigm for privacy-preserving data collection in distributed systems, where users contribute multi-dimensional records with potentially correlated attributes. Recent work has highlighted that correlation-induced privacy leakage (CPL) plays a critical role in shaping the privacy-utility trade-off under LDP, especially when correlations exist among attributes. Nevertheless, it remains unclear to what extent the prevailing assumptions and proposed solutions are valid and how significant CPL is in real-world data. To address this gap, we first perform a comprehensive statistical analysis of five widely used LDP mechanisms -- GRR, RAPPOR, OUE, OLH and Exponential mechanism -- to assess CPL across four real-world datasets. We identify that many primary assumptions and metrics in current approaches fall short of accurately characterising these leakages. Moreover, current studies have been limited to a set of pure LDP (i.e., {\delta = 0}) mechanisms. In response, we develop the first algorithmic framework to theoretically quantify CPL for any general approximated LDP (({\varepsilon},{\delta})-LDP) mechanism. We validate our theoretical results against empirical statistical results and provide a theoretical explanation for the observed statistical patterns. Finally, we propose two novel benchmarks to validate correlation analysis algorithms and evaluate the utility vs CPL of LDP mechanisms. Further, we demonstrate how these findings can be applied to achieve an efficient privacy-utility trade-off in real-world data governance.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 00:34:04 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Jayawardana",
"Sandaru",
""
],
[
"Ulukus",
"Sennur",
""
],
[
"Ding",
"Ming",
""
],
[
"Thilakarathna",
"Kanchana",
""
]
] |
2508.12561
|
Xuan Wang
|
X. Wang, H. J. Ong, S. Terashima, I. Tanihata, Y. K. Tanaka, N. Aoi, Y. Ayyad, J. Benlliure, F. Farinon, H. Fujioka, H. Geissel, J. Gellanki, C. L. Guo, E. Haettner, W. L. Hai, M. N. Harakeh, C. Hornung, K. Itahashi, R. Janik, N. Kalantar-Nayestanaki, R. Knobel, N. Kurz, K. Miki, I. Mukha, T. Myo, T. Nishi, D. Y. Pang, S. Pietri, A. Prochazka, C. Rappold, M. P. Reiter, J. L. Rodriguez-Sanchez, C. Scheidenberger, H. Simon, B. Sitar, P. Strmen, B. H. Sun, K. Suzuki, I. Szarka, H. Toki, H. Weick, E. Widmann, J. S. Winfield, X. D Xu, J. W. Zhao
|
Observation of Tensor-Driven High-Momentum Neutrons in ${}^{16}$O via ($p,d$) Reactions and Zero-Degree Deuteron Momentum Spectroscopy
| null | null | null | null |
nucl-ex
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The $^{16}\mathrm{O}(p,d)^{15}\mathrm{O}$ reaction has been studied at $0^{\circ}$ using 403-, 604-, 907- and 1209-MeV protons, comparing cross sections populating positive- and negative-parity states in $^{15}\mathrm{O}$. Transitions to positive-parity states exhibit strong sensitivity to high-momentum neutrons, while negative-parity transitions show much smaller effects. The cross-section ratio between positive- and negative-parity states rises sharply with momentum transfer, matching theoretical predictions that include tensor interactions, particularly the peak near $2~\mathrm{fm}^{-1}$ for the $5/2^{+}$ to ground-state ratio. These results highlight $0^{\circ}$ neutron-pickup reactions as a sensitive probe for tensor-driven high-momentum components, paving the way for studies in exotic nuclei via radioactive beams.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 01:40:27 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 10:42:57 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Wang",
"X.",
""
],
[
"Ong",
"H. J.",
""
],
[
"Terashima",
"S.",
""
],
[
"Tanihata",
"I.",
""
],
[
"Tanaka",
"Y. K.",
""
],
[
"Aoi",
"N.",
""
],
[
"Ayyad",
"Y.",
""
],
[
"Benlliure",
"J.",
""
],
[
"Farinon",
"F.",
""
],
[
"Fujioka",
"H.",
""
],
[
"Geissel",
"H.",
""
],
[
"Gellanki",
"J.",
""
],
[
"Guo",
"C. L.",
""
],
[
"Haettner",
"E.",
""
],
[
"Hai",
"W. L.",
""
],
[
"Harakeh",
"M. N.",
""
],
[
"Hornung",
"C.",
""
],
[
"Itahashi",
"K.",
""
],
[
"Janik",
"R.",
""
],
[
"Kalantar-Nayestanaki",
"N.",
""
],
[
"Knobel",
"R.",
""
],
[
"Kurz",
"N.",
""
],
[
"Miki",
"K.",
""
],
[
"Mukha",
"I.",
""
],
[
"Myo",
"T.",
""
],
[
"Nishi",
"T.",
""
],
[
"Pang",
"D. Y.",
""
],
[
"Pietri",
"S.",
""
],
[
"Prochazka",
"A.",
""
],
[
"Rappold",
"C.",
""
],
[
"Reiter",
"M. P.",
""
],
[
"Rodriguez-Sanchez",
"J. L.",
""
],
[
"Scheidenberger",
"C.",
""
],
[
"Simon",
"H.",
""
],
[
"Sitar",
"B.",
""
],
[
"Strmen",
"P.",
""
],
[
"Sun",
"B. H.",
""
],
[
"Suzuki",
"K.",
""
],
[
"Szarka",
"I.",
""
],
[
"Toki",
"H.",
""
],
[
"Weick",
"H.",
""
],
[
"Widmann",
"E.",
""
],
[
"Winfield",
"J. S.",
""
],
[
"Xu",
"X. D",
""
],
[
"Zhao",
"J. W.",
""
]
] |
2508.12601
|
Thomas R. Geballe
|
Yvonne J. Pendleton, T. R. Geballe, Laurie E. U. Chu, Marjorie Decleir, Karl D. Gordon, A. G. G. M. Tielens, Louis J. Allamandola, Jeroen Bouwman, J. E. Chiar, Curtis Dewitt, Burcu Gunay, Thomas Henning, Vito Mennella, M. E. Palumbo, Alexey Potapov, Maisie Rashman and Sascha Zeegers
|
A Tale of Two Sightlines: Comparison of Hydrocarbon Dust Absorption Bands toward Cygnus OB2-12 and the Galactic Center
|
28 pages, 7 figures, accepted by ApJ
| null | null | null |
astro-ph.GA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Infrared spectra of hydrocarbon dust absorption bands toward the bright hypergiant Cygnus OB2-12 are compared to published spectra of the Quintuplet Cluster, a sightline to the Galactic center. The Cyg OB2-12 data include a new ground-based 2.86-3.70 microns spectrum and a previously published, but here further analyzed, spectrum of the 5.50-7.34 microns region. Higher spectral resolution data for the Cyg OB2-12 sightline in the 3 micron region allows a detailed comparison of the 3.4 micron aliphatic bands to those observed toward the Quintuplet. Despite differences in interstellar environments along each sightline, strong similarities are observed in the central wavelengths and relative strengths for bands at 3.3, 3.4, 5.85, 6.2, and 6.85 microns. Analysis of these bands, produced by aromatic, aliphatic, olefinic, hydrogenated, and oxygenated components, shows that carbonaceous dust is a significant component of the diffuse interstellar medium, second in abundance only to silicates, and is primarily aromatic in nature. The grains producing these bands likely consist of large aromatic carbon cores with thin aliphatic mantles composed of hydrogenated amorphous carbon (HAC). Laboratory analog spectra reproduce the observed aliphatic absorption bands well, supporting the presence of such mantles. We present evidence that the carriers of both the 3.4 micron aliphatic and the 3.3 micron aromatic bands reside exclusively in the diffuse ISM, and that the 3.3 micron bands observed in the diffuse ISM differ from those seen in dense clouds, implying chemically distinct carriers.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 03:34:25 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 16:46:36 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Pendleton",
"Yvonne J.",
""
],
[
"Geballe",
"T. R.",
""
],
[
"Chu",
"Laurie E. U.",
""
],
[
"Decleir",
"Marjorie",
""
],
[
"Gordon",
"Karl D.",
""
],
[
"Tielens",
"A. G. G. M.",
""
],
[
"Allamandola",
"Louis J.",
""
],
[
"Bouwman",
"Jeroen",
""
],
[
"Chiar",
"J. E.",
""
],
[
"Dewitt",
"Curtis",
""
],
[
"Gunay",
"Burcu",
""
],
[
"Henning",
"Thomas",
""
],
[
"Mennella",
"Vito",
""
],
[
"Palumbo",
"M. E.",
""
],
[
"Potapov",
"Alexey",
""
],
[
"Rashman",
"Maisie",
""
],
[
"Zeegers",
"Sascha",
""
]
] |
2508.12615
|
Wenhao Zhang
|
Wenhao Zhang, Hao Zhu, Delong Wu, Di Kang, Linchao Bao, Xun Cao, Zhan Ma
|
WIPES: Wavelet-based Visual Primitives
|
IEEE/CVF International Conference on Computer Vision 2025
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Pursuing a continuous visual representation that offers flexible frequency modulation and fast rendering speed has recently garnered increasing attention in the fields of 3D vision and graphics. However, existing representations often rely on frequency guidance or complex neural network decoding, leading to spectrum loss or slow rendering. To address these limitations, we propose WIPES, a universal Wavelet-based vIsual PrimitivES for representing multi-dimensional visual signals. Building on the spatial-frequency localization advantages of wavelets, WIPES effectively captures both the low-frequency "forest" and the high-frequency "trees." Additionally, we develop a wavelet-based differentiable rasterizer to achieve fast visual rendering. Experimental results on various visual tasks, including 2D image representation, 5D static and 6D dynamic novel view synthesis, demonstrate that WIPES, as a visual primitive, offers higher rendering quality and faster inference than INR-based methods, and outperforms Gaussian-based representations in rendering quality.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 04:24:01 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 07:34:11 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Zhang",
"Wenhao",
""
],
[
"Zhu",
"Hao",
""
],
[
"Wu",
"Delong",
""
],
[
"Kang",
"Di",
""
],
[
"Bao",
"Linchao",
""
],
[
"Cao",
"Xun",
""
],
[
"Ma",
"Zhan",
""
]
] |
2508.12649
|
Lei Chen
|
Lei Chen, Michele Lanza, and Shinpei Hayashi
|
ChangePrism: Visualizing the Essence of Code Changes
|
(C) 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works
| null | null | null |
cs.SE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Understanding the changes made by developers when they submit a pull request and/or perform a commit on a repository is a crucial activity in software maintenance and evolution. The common way to review changes relies on examining code diffs, where textual differences between two file versions are highlighted in red and green to indicate additions and deletions of lines. This can be cumbersome for developers, making it difficult to obtain a comprehensive overview of all changes in a commit. Moreover, certain types of code changes can be particularly significant and may warrant differentiation from standard modifications to enhance code comprehension. We present a novel visualization approach supported by a tool named ChangePrism, which provides a way to better understand code changes. The tool comprises two components: extraction, which retrieves code changes and relevant information from the git history, and visualization, which offers both general and detailed views of code changes in commits. The general view provides an overview of different types of code changes across commits, while the detailed view displays the exact changes in the source code for each commit.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 06:23:34 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 13:50:27 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Chen",
"Lei",
""
],
[
"Lanza",
"Michele",
""
],
[
"Hayashi",
"Shinpei",
""
]
] |
2508.12672
|
Emmanouil Kritharakis
|
Emmanouil Kritharakis, Dusan Jakovetic, Antonios Makris, Konstantinos Tserpes
|
Robust Federated Learning under Adversarial Attacks via Loss-Based Client Clustering
|
16 pages, 5 figures
| null | null | null |
cs.LG cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Federated Learning (FL) enables collaborative model training across multiple clients without sharing private data. We consider FL scenarios wherein FL clients are subject to adversarial (Byzantine) attacks, while the FL server is trusted (honest) and has a trustworthy side dataset. This may correspond to, e.g., cases where the server possesses trusted data prior to federation, or to the presence of a trusted client that temporarily assumes the server role. Our approach requires only two honest participants, i.e., the server and one client, to function effectively, without prior knowledge of the number of malicious clients. Theoretical analysis demonstrates bounded optimality gaps even under strong Byzantine attacks. Experimental results show that our algorithm significantly outperforms standard and robust FL baselines such as Mean, Trimmed Mean, Median, Krum, and Multi-Krum under various attack strategies including label flipping, sign flipping, and Gaussian noise addition across MNIST, FMNIST, and CIFAR-10 benchmarks using the Flower framework.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 07:11:21 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 05:06:11 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Kritharakis",
"Emmanouil",
""
],
[
"Jakovetic",
"Dusan",
""
],
[
"Makris",
"Antonios",
""
],
[
"Tserpes",
"Konstantinos",
""
]
] |
2508.12681
|
Johann Licher
|
Johann Licher, Max Bartholdt, Henrik Krauss, Tim-Lukas Habich, Thomas Seel, Moritz Schappler
|
Adaptive Model-Predictive Control of a Soft Continuum Robot Using a Physics-Informed Neural Network Based on Cosserat Rod Theory
|
20 pages, 15 figures
| null | null | null |
cs.RO cs.LG cs.SY eess.SY
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Dynamic control of soft continuum robots (SCRs) holds great potential for expanding their applications, but remains a challenging problem due to the high computational demands of accurate dynamic models. While data-driven approaches like Koopman-operator-based methods have been proposed, they typically lack adaptability and cannot capture the full robot shape, limiting their applicability. This work introduces a real-time-capable nonlinear model-predictive control (MPC) framework for SCRs based on a domain-decoupled physics-informed neural network (DD-PINN) with adaptable bending stiffness. The DD-PINN serves as a surrogate for the dynamic Cosserat rod model with a speed-up factor of 44000. It is also used within an unscented Kalman filter for estimating the model states and bending compliance from end-effector position measurements. We implement a nonlinear evolutionary MPC running at 70 Hz on the GPU. In simulation, it demonstrates accurate tracking of dynamic trajectories and setpoint control with end-effector position errors below 3 mm (2.3% of the actuator's length). In real-world experiments, the controller achieves similar accuracy and accelerations up to 3.55 m/s2.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 07:24:36 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Licher",
"Johann",
""
],
[
"Bartholdt",
"Max",
""
],
[
"Krauss",
"Henrik",
""
],
[
"Habich",
"Tim-Lukas",
""
],
[
"Seel",
"Thomas",
""
],
[
"Schappler",
"Moritz",
""
]
] |
2508.12686
|
Jun He Prof.
|
Zuo-Ming Ding, Qi Huang, Jun He
|
Roles of $\bar{D}^{*}K^{*}$ and $D^*\bar{D}$ molecular states in decay $B^+ \to D^{*+} D^- K^+$
|
12 pages, 10 figures
| null | null | null |
hep-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This study investigates the three-body decay process $B^+ \to D^{*+} D^- K^+$, aiming to explore the possible origins of $T^*_{\bar{c}\bar{s}0}(2870)^0$ and $\chi_{c1}(3872)$ as intermediate states. Within the molecular state framework, $T^*_{\bar{c}\bar{s}0}(2870)^0$ and $\chi_{c1}(3872)$ are considered as possible $\bar{D}^{*}K^{}$ and $D^*\bar{D}$ molecular states, respectively. Using effective Lagrangians, the interaction kernels of the $\bar{D}^{*}K^{*}$ and $D^*\bar{D}$ systems are constructed within the one-boson-exchange model. The corresponding rescattering amplitudes and pole positions are obtained by solving the quasipotential Bethe-Salpeter equation. These amplitudes are incorporated into the decay amplitude of the three-body process, and the $D^-K^+$ and $D^{*+}D^-$ invariant mass spectra are simulated via Monte Carlo methods. To better reproduce the experimental data, additional Breit-Wigner contributions from $T^*_{\bar{c}\bar{s}1}(2900)^0$, $\chi_{c1}(4010)$, and $h_c(4300)$ are included. The results show a pronounced enhancement near 2900 MeV in the $D^-K^+$ invariant mass spectrum, strongly supporting the interpretation of $T^*_{\bar{c}\bar{s}0}(2870)^0$ as a $\bar{D}^{*}K^{*}$ molecular state. While the $\bar{D}^{*}K^{*}$ molecular state provides a reasonable contribution to the $D^-K^+$ spectrum, the $D^*\bar{D}$ molecular state yields no significant effect on either the $D^-K^+$ or $D^{*+}D^-$ distributions. This suggests that the observed $\chi_{c1}(3872)$ structure around 3872 MeV may not be interpreted as a $D^*\bar{D}$ molecular state.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 07:39:51 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 13:33:41 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Ding",
"Zuo-Ming",
""
],
[
"Huang",
"Qi",
""
],
[
"He",
"Jun",
""
]
] |
2508.12734
|
Min Liu
|
Elnur Emrah and Patrik L. Ferrari and Min Liu
|
Critical fluctuations of last passage percolation with thick boundaries
|
44 pages, 5 figures
| null | null | null |
math.PR
|
http://creativecommons.org/licenses/by/4.0/
|
We consider the exponential last passage percolation (LPP) with thick two-sided boundary that consists of a few inhomogeneous columns and rows. Ben Arous and Corwin previously studied the limit fluctuations in this model except in a critical regime, for which they predicted that the limit distribution exists and depends only on the most dominant columns and rows. In this article, we prove their conjecture and identify the limit distribution explicitly in terms of a Fredholm determinant of a $2 \times 2$ matrix kernel. This result leads in particular to an explicit variational formula for the one-point marginal of the KPZ fixed point for a new class of initial conditions. Our limit distribution is also a novel generalization of and provides a new numerically efficient representation for the Baik--Rains distribution.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 09:00:14 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 07:14:58 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Emrah",
"Elnur",
""
],
[
"Ferrari",
"Patrik L.",
""
],
[
"Liu",
"Min",
""
]
] |
2508.12738
|
Rolf Findeisen
|
Sebastian Hirt, Lukas Theiner, Maik Pfefferkorn, Rolf Findeisen
|
A Hierarchical Surrogate Model for Efficient Multi-Task Parameter Learning in Closed-Loop Control
|
8 pages, 4 figures, accepted for CDC 2025
| null | null | null |
eess.SY cs.LG cs.SY
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Many control problems require repeated tuning and adaptation of controllers across distinct closed-loop tasks, where data efficiency and adaptability are critical. We propose a hierarchical Bayesian optimization (BO) framework that is tailored to efficient controller parameter learning in sequential decision-making and control scenarios for distinct tasks. Instead of treating the closed-loop cost as a black-box, our method exploits structural knowledge of the underlying problem, consisting of a dynamical system, a control law, and an associated closed-loop cost function. We construct a hierarchical surrogate model using Gaussian processes that capture the closed-loop state evolution under different parameterizations, while the task-specific weighting and accumulation into the closed-loop cost are computed exactly via known closed-form expressions. This allows knowledge transfer and enhanced data efficiency between different closed-loop tasks. The proposed framework retains sublinear regret guarantees on par with standard black-box BO, while enabling multi-task or transfer learning. Simulation experiments with model predictive control demonstrate substantial benefits in both sample efficiency and adaptability when compared to purely black-box BO approaches.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 09:01:28 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 06:46:05 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Hirt",
"Sebastian",
""
],
[
"Theiner",
"Lukas",
""
],
[
"Pfefferkorn",
"Maik",
""
],
[
"Findeisen",
"Rolf",
""
]
] |
2508.12794
|
Kyriaki Kokka
|
Kyriaki (Kelly) Kokka, Rahul Goel, Ali Abbas, Kerry A. Nice, Luca Martial, SM Labib, Rihuan Ke, Carola Bibiane Sch\"onlieb, James Woodcock
|
Vehicle detection from GSV imagery: Predicting travel behaviour for cycling and motorcycling using Computer Vision
| null | null | null | null |
cs.CV cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Transportation influence health by shaping exposure to physical activity, air pollution and injury risk. Comparative data on cycling and motorcycling behaviours is scarce, particularly at a global scale. Street view imagery, such as Google Street View (GSV), combined with computer vision, is a valuable resource for efficiently capturing travel behaviour data. This study demonstrates a novel approach using deep learning on street view images to estimate cycling and motorcycling levels across diverse cities worldwide. We utilized data from 185 global cities. The data on mode shares of cycling and motorcycling estimated using travel surveys or censuses. We used GSV images to detect cycles and motorcycles in sampled locations, using 8000 images per city. The YOLOv4 model, fine-tuned using images from six cities, achieved a mean average precision of 89% for detecting cycles and motorcycles. A global prediction model was developed using beta regression with city-level mode shares as outcome, with log transformed explanatory variables of counts of GSV-detected images with cycles and motorcycles, while controlling for population density. We found strong correlations between GSV motorcycle counts and motorcycle mode share (0.78) and moderate correlations between GSV cycle counts and cycling mode share (0.51). Beta regression models predicted mode shares with $R^2$ values of 0.614 for cycling and 0.612 for motorcycling, achieving median absolute errors (MDAE) of 1.3% and 1.4%, respectively. Scatterplots demonstrated consistent prediction accuracy, though cities like Utrecht and Cali were outliers. The model was applied to 60 cities globally for which we didn't have recent mode share data. We provided estimates for some cities in the Middle East, Latin America and East Asia. With computer vision, GSV images capture travel modes and activity, providing insights alongside traditional data sources.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 10:17:30 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 14:43:20 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Kyriaki",
"",
"",
"Kelly"
],
[
"Kokka",
"",
""
],
[
"Goel",
"Rahul",
""
],
[
"Abbas",
"Ali",
""
],
[
"Nice",
"Kerry A.",
""
],
[
"Martial",
"Luca",
""
],
[
"Labib",
"SM",
""
],
[
"Ke",
"Rihuan",
""
],
[
"Schönlieb",
"Carola Bibiane",
""
],
[
"Woodcock",
"James",
""
]
] |
2508.12800
|
Dai Yuqin
|
Yong Deng, Guoqing Wang, Zhenzhe Ying, Xiaofeng Wu, Jinzhen Lin, Wenwen Xiong, Yuqin Dai, Shuo Yang, Zhanwei Zhang, Qiwen Wang, Yang Qin, Changhua Meng
|
Atom-Searcher: Enhancing Agentic Deep Research via Fine-Grained Atomic Thought Reward
| null | null | null | null |
cs.CL cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Large language models (LLMs) exhibit remarkable problem-solving abilities, but struggle with complex tasks due to static internal knowledge. Retrieval-Augmented Generation (RAG) enhances access to external information, yet remains limited in multi-hop reasoning and strategic search due to rigid workflows. Recent advancements in agentic deep research empower LLMs to autonomously reason, search, and synthesize information. However, current approaches relying on outcome-based reinforcement learning (RL) face critical issues such as conflicting gradients and reward sparsity, limiting performance gains and training efficiency. To address these, we first propose Atomic Thought, a novel LLM thinking paradigm that decomposes reasoning into fine-grained functional units. These units are supervised by Reasoning Reward Models (RRMs), which provide Atomic Thought Rewards (ATR) for fine-grained guidance. Building on this, we propose Atom-Searcher, a novel RL framework for agentic deep research that integrates Atomic Thought and ATR. Atom-Searcher uses a curriculum-inspired reward schedule, prioritizing process-level ATR early and transitioning to outcome rewards, accelerating convergence on effective reasoning paths. Experiments on seven benchmarks show consistent improvements over the state-of-the-art. Key advantages include: (1) Atom-Searcher scales computation at test-time. (2) Atomic Thought provides supervision anchors for RRMs, bridging deep research tasks and RRMs. (3) Atom-Searcher exhibits more interpretable, human-like reasoning patterns.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 10:23:10 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 11:40:33 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Deng",
"Yong",
""
],
[
"Wang",
"Guoqing",
""
],
[
"Ying",
"Zhenzhe",
""
],
[
"Wu",
"Xiaofeng",
""
],
[
"Lin",
"Jinzhen",
""
],
[
"Xiong",
"Wenwen",
""
],
[
"Dai",
"Yuqin",
""
],
[
"Yang",
"Shuo",
""
],
[
"Zhang",
"Zhanwei",
""
],
[
"Wang",
"Qiwen",
""
],
[
"Qin",
"Yang",
""
],
[
"Meng",
"Changhua",
""
]
] |
2508.12832
|
Jinyu Lu
|
Jinyu Lu, Xinrong Sun, Yunting Tao, Tong Ji, Fanyu Kong, Guoqiang Yang
|
Efficient and Verifiable Privacy-Preserving Convolutional Computation for CNN Inference with Untrusted Clouds
|
Conference link: [ICIC 2025](http://www.ic-icc.cn/2025/index.php) will provide further details
|
International Conference On Intelligent Computing 2025, Ningbo, China, July 26-29, 2025, Volume I, pp. 866-881
| null | null |
cs.CR cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The widespread adoption of convolutional neural networks (CNNs) in resource-constrained scenarios has driven the development of Machine Learning as a Service (MLaaS) system. However, this approach is susceptible to privacy leakage, as the data sent from the client to the untrusted cloud server often contains sensitive information. Existing CNN privacy-preserving schemes, while effective in ensuring data confidentiality through homomorphic encryption and secret sharing, face efficiency bottlenecks, particularly in convolution operations. In this paper, we propose a novel verifiable privacy-preserving scheme tailored for CNN convolutional layers. Our scheme enables efficient encryption and decryption, allowing resource-constrained clients to securely offload computations to the untrusted cloud server. Additionally, we present a verification mechanism capable of detecting the correctness of the results with a success probability of at least $1-\frac{1}{\left|Z\right|}$. Extensive experiments conducted on 10 datasets and various CNN models demonstrate that our scheme achieves speedups ranging $26 \times$ ~ $\ 87\times$ compared to the original plaintext model while maintaining accuracy.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 11:17:53 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 07:03:27 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Lu",
"Jinyu",
""
],
[
"Sun",
"Xinrong",
""
],
[
"Tao",
"Yunting",
""
],
[
"Ji",
"Tong",
""
],
[
"Kong",
"Fanyu",
""
],
[
"Yang",
"Guoqiang",
""
]
] |
2508.12837
|
Gizem Y\"uce
|
Aditya Varre, Gizem Y\"uce, Nicolas Flammarion
|
Learning In-context n-grams with Transformers: Sub-n-grams Are Near-stationary Points
|
ICML2025
| null | null | null |
cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Motivated by empirical observations of prolonged plateaus and stage-wise progression during training, we investigate the loss landscape of transformer models trained on in-context next-token prediction tasks. In particular, we focus on learning in-context $n$-gram language models under cross-entropy loss, and establish a sufficient condition for parameter configurations to be stationary points. We then construct a set of parameter configurations for a simplified transformer model that represent $k$-gram estimators (for $k \leq n$), and show that the gradient of the population loss at these solutions vanishes in the limit of infinite sequence length and parameter norm. This reveals a key property of the loss landscape: {sub-$n$-grams are near-stationary points of the population cross-entropy loss}, offering theoretical insight into widely observed phenomena such as stage-wise learning dynamics and emergent phase transitions. These insights are further supported by numerical experiments that illustrate the learning dynamics of $n$-grams, characterized by discrete transitions between near-stationary solutions.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 11:24:30 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 09:36:39 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Varre",
"Aditya",
""
],
[
"Yüce",
"Gizem",
""
],
[
"Flammarion",
"Nicolas",
""
]
] |
2508.12839
|
Tiancheng Zhang
|
Tiancheng Zhang, Cheng Zhang, Shuren Liu, Xiaofei Wang, Shaoyuan Huang, Wenyu Wang
|
HRS: Hybrid Representation Framework with Scheduling Awareness for Time Series Forecasting in Crowdsourced Cloud-Edge Platforms
|
10 pages, 14 figures, ECAI2025
| null | null | null |
cs.LG cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
With the rapid proliferation of streaming services, network load exhibits highly time-varying and bursty behavior, posing serious challenges for maintaining Quality of Service (QoS) in Crowdsourced Cloud-Edge Platforms (CCPs). While CCPs leverage Predict-then-Schedule architecture to improve QoS and profitability, accurate load forecasting remains challenging under traffic surges. Existing methods either minimize mean absolute error, resulting in underprovisioning and potential Service Level Agreement (SLA) violations during peak periods, or adopt conservative overprovisioning strategies, which mitigate SLA risks at the expense of increased resource expenditure. To address this dilemma, we propose HRS, a hybrid representation framework with scheduling awareness that integrates numerical and image-based representations to better capture extreme load dynamics. We further introduce a Scheduling-Aware Loss (SAL) that captures the asymmetric impact of prediction errors, guiding predictions that better support scheduling decisions. Extensive experiments on four real-world datasets demonstrate that HRS consistently outperforms ten baselines and achieves state-of-the-art performance, reducing SLA violation rates by 63.1% and total profit loss by 32.3%.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 11:25:54 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 11:02:50 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Zhang",
"Tiancheng",
""
],
[
"Zhang",
"Cheng",
""
],
[
"Liu",
"Shuren",
""
],
[
"Wang",
"Xiaofei",
""
],
[
"Huang",
"Shaoyuan",
""
],
[
"Wang",
"Wenyu",
""
]
] |
2508.12914
|
Brad Turow
|
Brad Turow and Jose A. Perea
|
Discrete Approximate Circle Bundles
| null | null | null | null |
math.AT
|
http://creativecommons.org/licenses/by/4.0/
|
In this paper, we introduce discrete approximate circle bundles, a class of objects designed to serve as the data science analog of circle bundles from algebraic topology. We show that, under appropriate conditions, one can meaningfully and stably identify a discrete approximate circle bundle with an isomorphism class of true circle bundles. We also describe two cohomology invariants which uniquely determine the isomorphism class of a circle bundle, and provide algorithms to compute them given a discrete approximate representative. Finally, we propose a novel methodology for coordinatization and dimensionality reduction of circle bundle data. To illustrate the practical utility and viability of our algorithms, we present applications to both real and synthetic datasets from computer vision (e.g., modeling optical flow).
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 13:22:43 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 02:13:31 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Turow",
"Brad",
""
],
[
"Perea",
"Jose A.",
""
]
] |
2508.12915
|
Baojun Fang
|
Bruce Fang and Steven J. Miller
|
Benford behavior resulting from stick and box fragmentation processes
|
33 pages and 0 figures
| null | null | null |
math.PR
|
http://creativecommons.org/licenses/by/4.0/
|
Benford's law is the statement that in many real world data sets, the probability of having digit $d$ in base $B$, where $1 \leq d \leq B$, as the first digit is \log_{B}\!\left(\frac{d+1}{d}\right). We sometimes refer to this as weak Benford behavior, and we say that a data set exhibits strong Benford behavior in base $B$ if the probability of having significand at most s, where $1 \leq s < B$, is \log_{B}\!\left(s\right). We examine Benford behaviors in two different probabilistic models: stick and box fragmentation. Building on the work by arXiv:1309.5603 on the one-dimensional stick fragmentation model, we employ combinatorial identities on multinomial coefficients to reduce the high-dimensional stick fragmentation model to the one-dimensional model and provide a necessary and sufficient condition for the lengths of the stick fragments to converge to strong Benford behavior. Then we answer a conjecture by arXiv:2304.08335 on the high-dimensional box fragmentation model. Using tools from order statistics, we prove that under some mild conditions, faces of any arbitrary dimension of the box have total volume converging to strong Benford behavior.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 13:23:06 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 01:55:54 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Fang",
"Bruce",
""
],
[
"Miller",
"Steven J.",
""
]
] |
2508.12942
|
Kyriaki-Margarita Bintsi
|
Kyriaki-Margarita Bintsi, Ya\"el Balbastre, Jingjing Wu, Julia F. Lehman, Suzanne N. Haber, and Anastasia Yendiki
|
Fully Automated Segmentation of Fiber Bundles in Anatomic Tracing Data
|
Accepted at CDMRI, MICCAI 2025
| null | null | null |
cs.CV cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Anatomic tracer studies are critical for validating and improving diffusion MRI (dMRI) tractography. However, large-scale analysis of data from such studies is hampered by the labor-intensive process of annotating fiber bundles manually on histological slides. Existing automated methods often miss sparse bundles or require complex post-processing across consecutive sections, limiting their flexibility and generalizability. We present a streamlined, fully automated framework for fiber bundle segmentation in macaque tracer data, based on a U-Net architecture with large patch sizes, foreground aware sampling, and semisupervised pre-training. Our approach eliminates common errors such as mislabeling terminals as bundles, improves detection of sparse bundles by over 20% and reduces the False Discovery Rate (FDR) by 40% compared to the state-of-the-art, all while enabling analysis of standalone slices. This new framework will facilitate the automated analysis of anatomic tracing data at a large scale, generating more ground-truth data that can be used to validate and optimize dMRI tractography methods.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 14:17:24 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 15:58:58 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Bintsi",
"Kyriaki-Margarita",
""
],
[
"Balbastre",
"Yaël",
""
],
[
"Wu",
"Jingjing",
""
],
[
"Lehman",
"Julia F.",
""
],
[
"Haber",
"Suzanne N.",
""
],
[
"Yendiki",
"Anastasia",
""
]
] |
2508.12946
|
Stefan Schiffer
|
Ann-Sophie L. Schenk and Stefan Schiffer and Heqiu Song
|
Insights from Interviews with Teachers and Students on the Use of a Social Robot in Computer Science Class in Sixth Grade
|
4 pages, 2 figures, Late Breaking Report accepted for RO-MAN 2025
| null | null | null |
cs.RO cs.HC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper we report on first insights from interviews with teachers and students on using social robots in computer science class in sixth grade. Our focus is on learning about requirements and potential applications. We are particularly interested in getting both perspectives, the teachers' and the learners' view on how robots could be used and what features they should or should not have. Results show that teachers as well as students are very open to robots in the classroom. However, requirements are partially quite heterogeneous among the groups. This leads to complex design challenges which we discuss at the end of this paper.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 14:22:28 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 11:08:27 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Schenk",
"Ann-Sophie L.",
""
],
[
"Schiffer",
"Stefan",
""
],
[
"Song",
"Heqiu",
""
]
] |
2508.12992
|
Xiangxian Li
|
Xiangxian Li, Yawen Zheng, Baiqiao Zhang, Yijia Ma, Xianhui Cao, Juan Liu, Yulong Bian, Jin Huang, Chenglei Yang
|
MAGNeT: Multimodal Adaptive Gaussian Networks for Intent Inference in Moving Target Selection across Complex Scenarios
|
Accepted by ACM MM 2025
| null | null | null |
cs.MM
|
http://creativecommons.org/licenses/by/4.0/
|
Moving target selection in multimedia interactive systems faces unprecedented challenges as users increasingly interact across diverse and dynamic contexts-from live streaming in moving vehicles to VR gaming in varying environments. Existing approaches rely on probabilistic models that relate endpoint distribution to target properties such as size and speed. However, these methods require substantial training data for each new context and lack transferability across scenarios, limiting their practical deployment in diverse multimedia environments where rich multimodal contextual information is readily available. This paper introduces MAGNeT (Multimodal Adaptive Gaussian Networks), which addresses these problems by combining classical statistical modeling with a context-aware multimodal method. MAGNeT dynamically fuses pre-fitted Ternary-Gaussian models from various scenarios based on real-time contextual cues, enabling effective adaptation with minimal training data while preserving model interpretability. We conduct experiments on self-constructed 2D and 3D moving target selection datasets under in-vehicle vibration conditions. Extensive experiments demonstrate that MAGNeT achieves lower error rates with few-shot samples by applying context-aware fusion of Gaussian experts from multi-factor conditions.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 15:12:27 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 09:08:17 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Li",
"Xiangxian",
""
],
[
"Zheng",
"Yawen",
""
],
[
"Zhang",
"Baiqiao",
""
],
[
"Ma",
"Yijia",
""
],
[
"Cao",
"Xianhui",
""
],
[
"Liu",
"Juan",
""
],
[
"Bian",
"Yulong",
""
],
[
"Huang",
"Jin",
""
],
[
"Yang",
"Chenglei",
""
]
] |
2508.13021
|
Pengcheng Huang
|
Pengcheng Huang, Shuhao Liu, Zhenghao Liu, Yukun Yan, Shuo Wang, Zulong Chen, Tong Xiao
|
PC-Sampler: Position-Aware Calibration of Decoding Bias in Masked Diffusion Models
|
17 pages,13 figures
| null | null | null |
cs.AI cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent advances in masked diffusion models (MDMs) have established them as powerful non-autoregressive alternatives for sequence generation. Nevertheless, our preliminary experiments reveal that the generation quality of MDMs is still highly sensitive to the choice of decoding strategy. In particular, widely adopted uncertainty-based samplers suffer from two key limitations: a lack of global trajectory control and a pronounced bias toward trivial tokens in the early stages of decoding. These shortcomings restrict the full potential of MDMs. In this work, we introduce Position-Aware Confidence-Calibrated Sampling (PC-Sampler), a novel decoding strategy that unifies global trajectory planning with content-aware informativeness maximization. PC-Sampler incorporates a position-aware weighting mechanism to regulate the decoding path and a calibrated confidence score to suppress the premature selection of trivial tokens. Extensive experiments on three advanced MDMs across seven challenging benchmarks-including logical reasoning and planning tasks-demonstrate that PC-Sampler consistently outperforms existing MDM decoding strategies by more than 10% on average, significantly narrowing the performance gap with state-of-the-art autoregressive models. All codes are available at https://github.com/NEUIR/PC-Sampler.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 15:38:37 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 02:03:36 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Huang",
"Pengcheng",
""
],
[
"Liu",
"Shuhao",
""
],
[
"Liu",
"Zhenghao",
""
],
[
"Yan",
"Yukun",
""
],
[
"Wang",
"Shuo",
""
],
[
"Chen",
"Zulong",
""
],
[
"Xiao",
"Tong",
""
]
] |
2508.13025
|
Tiago Duarte Guerreiro
|
Tiago Duarte Guerreiro, Takuzo Okada
|
A survey on birational Rigidity of threefold Weighted Complete Intersections
|
37 pp. Comments welcome
| null | null | null |
math.AG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We survey what is known about Fano threefold weighted complete intersections from the point of view of birational rigidity.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 15:41:57 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 08:07:25 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Guerreiro",
"Tiago Duarte",
""
],
[
"Okada",
"Takuzo",
""
]
] |
2508.13032
|
Remy El Sabeh
|
Nicolas Bousquet, Remy El Sabeh, Amer E. Mouawad, Naomi Nishimura
|
On the complexity of constrained reconfiguration and motion planning
|
Looking to incorporate comments from reviewers
| null | null | null |
cs.CC cs.DM cs.DS cs.RO math.CO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Coordinating the motion of multiple agents in constrained environments is a fundamental challenge in robotics, motion planning, and scheduling. A motivating example involves $n$ robotic arms, each represented as a line segment. The objective is to rotate each arm to its vertical orientation, one at a time (clockwise or counterclockwise), without collisions nor rotating any arm more than once. This scenario is an example of the more general $k$-Compatible Ordering problem, where $n$ agents, each capable of $k$ state-changing actions, must transition to specific target states under constraints encoded as a set $\mathcal{G}$ of $k$ pairs of directed graphs.
We show that $k$-Compatible Ordering is $\mathsf{NP}$-complete, even when $\mathcal{G}$ is planar, degenerate, or acyclic. On the positive side, we provide polynomial-time algorithms for cases such as when $k = 1$ or $\mathcal{G}$ has bounded treewidth. We also introduce generalized variants supporting multiple state-changing actions per agent, broadening the applicability of our framework. These results extend to a wide range of scheduling, reconfiguration, and motion planning applications in constrained environments.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 15:50:57 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 07:53:31 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Bousquet",
"Nicolas",
""
],
[
"Sabeh",
"Remy El",
""
],
[
"Mouawad",
"Amer E.",
""
],
[
"Nishimura",
"Naomi",
""
]
] |
2508.13062
|
Yu-Chen Wang
|
Yu-Chen Wang, Youhui Yun, Hong-Jian He, Yue Meng
|
Search for Light Inelastic Dark Matter with Low-Energy Ionization Signatures in PandaX-4T
|
6 pages, 1 ancillary file
| null | null | null |
hep-ph astro-ph.CO hep-ex
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Direct detection of light dark matter (DM) is generally difficult due to its small recoil energy. The inelastic scattering of DM can produce unique signatures in the DM direct detection experiments. Using the low-energy unpaired ionization data from PandaX-4T, we newly analyze the probe of the exothermic inelastic dark matter (ineDM). We demonstrate that PandaX-4T can probe the ineDM mass-splitting down to 0.05 keV and probe the ineDM mass to the sub-GeV range. For the ineDM with a dark photon mediator, we use the PandaX-4T data to impose stringent bounds on the mixing parameter between the dark photon and photon.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 16:33:24 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Wang",
"Yu-Chen",
""
],
[
"Yun",
"Youhui",
""
],
[
"He",
"Hong-Jian",
""
],
[
"Meng",
"Yue",
""
]
] |
2508.13128
|
Jules Colas Mr
|
J. Colas, A. Emmanuelli, D. Dragna, R.J.A.M. Stevens
|
Modeling wind farm noise emission and propagation: effects of flow and layout
|
Submitted to Renewable Energy
| null | null | null |
physics.flu-dyn physics.app-ph physics.comp-ph
|
http://creativecommons.org/licenses/by/4.0/
|
This study demonstrates how wind farm flow physics influence noise generation and downstream propagation through numerical simulations. The flow field is modeled using large-eddy simulations (LES), and the time-averaged output serves as input to acoustic models that predict wind turbine noise. In the first turbine row, turbulence-induced noise (TIN) and trailing edge noise (TEN) contribute equally, with TIN dominating at low frequencies and TEN at higher frequencies. Downstream, TEN decreases due to lower wind speeds, while TIN mostly persists due to increased turbulence dissipation. These effects are more pronounced in aligned wind farms, where stronger wake interactions occur, than in staggered layouts. However, staggered farms produce more noise overall because turbines operate at higher wind speeds.Additionally, wind farm flow significantly affects sound propagation downwind. The wake superposition modifies sound focusing leading to different amplification area than for an isolated turbine. For a staggered layout it particularly shows enhanced sound focusing downwind due to the position of the turbine wakes. This leads to higher sound levels and higher amplitude modulation downwind for the wind farm compared to an aligned layout. These phenomena are not captured by models based on isolated turbines. These findings underscore the importance of integrating flow and acoustic models to more accurately assess the environmental impact of wind farms.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 17:36:41 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 06:56:25 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Colas",
"J.",
""
],
[
"Emmanuelli",
"A.",
""
],
[
"Dragna",
"D.",
""
],
[
"Stevens",
"R. J. A. M.",
""
]
] |
2508.13135
|
Yueyang Liu
|
Yueyang Liu, Lance Kennedy, Ruochen Kong, Joon-Seok Kim, Andreas Z\"ufle
|
Training Machine Learning Models on Human Spatio-temporal Mobility Data: An Experimental Study [Experiment Paper]
|
This paper is the extended version of our work accepted at the 33rd ACM International Conference on Advances in Geographic Information Systems
| null |
10.1145/3748636.3762740
| null |
cs.LG
|
http://creativecommons.org/licenses/by-sa/4.0/
|
Individual-level human mobility prediction has emerged as a significant topic of research with applications in infectious disease monitoring, child, and elderly care. Existing studies predominantly focus on the microscopic aspects of human trajectories: such as predicting short-term trajectories or the next location visited, while offering limited attention to macro-level mobility patterns and the corresponding life routines. In this paper, we focus on an underexplored problem in human mobility prediction: determining the best practices to train a machine learning model using historical data to forecast an individuals complete trajectory over the next days and weeks. In this experiment paper, we undertake a comprehensive experimental analysis of diverse models, parameter configurations, and training strategies, accompanied by an in-depth examination of the statistical distribution inherent in human mobility patterns. Our empirical evaluations encompass both Long Short-Term Memory and Transformer-based architectures, and further investigate how incorporating individual life patterns can enhance the effectiveness of the prediction. We show that explicitly including semantic information such as day-of-the-week and user-specific historical information can help the model better understand individual patterns of life and improve predictions. Moreover, since the absence of explicit user information is often missing due to user privacy, we show that the sampling of users may exacerbate data skewness and result in a substantial loss in predictive accuracy. To mitigate data imbalance and preserve diversity, we apply user semantic clustering with stratified sampling to ensure that the sampled dataset remains representative. Our results further show that small-batch stochastic gradient optimization improves model performance, especially when human mobility training data is limited.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 17:49:10 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Aug 2025 01:52:50 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Liu",
"Yueyang",
""
],
[
"Kennedy",
"Lance",
""
],
[
"Kong",
"Ruochen",
""
],
[
"Kim",
"Joon-Seok",
""
],
[
"Züfle",
"Andreas",
""
]
] |
2508.13156
|
Ping Guo
|
Ping Guo, Yiting Wang, Wanghao Ye, Yexiao He, Ziyao Wang, Xiaopeng Dai, Ang Li, Qingfu Zhang
|
EvoVerilog: Large Langugage Model Assisted Evolution of Verilog Code
| null | null | null | null |
cs.AR cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Large Language Models (LLMs) have demonstrated great potential in automating the generation of Verilog hardware description language code for hardware design. This automation is critical to reducing human effort in the complex and error-prone process of hardware design.
However, existing approaches predominantly rely on human intervention and fine-tuning using curated datasets, limiting their scalability in automated design workflows.
Although recent iterative search techniques have emerged, they often fail to explore diverse design solutions and may underperform simpler approaches such as repeated prompting.
To address these limitations, we introduce EvoVerilog, a novel framework that combines the reasoning capabilities of LLMs with evolutionary algorithms to automatically generate and refine Verilog code.
EvoVerilog utilizes a multiobjective, population-based search strategy to explore a wide range of design possibilities without requiring human intervention.
Extensive experiments demonstrate that EvoVerilog achieves state-of-the-art performance, with pass@10 scores of 89.1 and 80.2 on the VerilogEval-Machine and VerilogEval-Human benchmarks, respectively. Furthermore, the framework showcases its ability to explore diverse designs by simultaneously generating a variety of functional Verilog code while optimizing resource utilization.
|
[
{
"version": "v1",
"created": "Thu, 26 Jun 2025 13:32:25 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Guo",
"Ping",
""
],
[
"Wang",
"Yiting",
""
],
[
"Ye",
"Wanghao",
""
],
[
"He",
"Yexiao",
""
],
[
"Wang",
"Ziyao",
""
],
[
"Dai",
"Xiaopeng",
""
],
[
"Li",
"Ang",
""
],
[
"Zhang",
"Qingfu",
""
]
] |
2508.13157
|
Xu Haohang
|
Haohang Xu, Chengjie Liu, Qihang Wang, Wenhao Huang, Yongjian Xu, Weiyu Chen, Anlan Peng, Zhijun Li, Bo Li, Lei Qi, Jun Yang, Yuan Du, and Li Du
|
Image2Net: Datasets, Benchmark and Hybrid Framework to Convert Analog Circuit Diagrams into Netlists
|
10 pages, 12 figures, 6 tables
| null | null | null |
cs.AR cs.AI cs.CV eess.IV
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Large Language Model (LLM) exhibits great potential in designing of analog integrated circuits (IC) because of its excellence in abstraction and generalization for knowledge. However, further development of LLM-based analog ICs heavily relies on textual description of analog ICs, while existing analog ICs are mostly illustrated in image-based circuit diagrams rather than text-based netlists. Converting circuit diagrams to netlists help LLMs to enrich the knowledge of analog IC. Nevertheless, previously proposed conversion frameworks face challenges in further application because of limited support of image styles and circuit elements. Up to now, it still remains a challenging task to effectively convert complex circuit diagrams into netlists. To this end, this paper constructs and opensources a new dataset with rich styles of circuit diagrams as well as balanced distribution of simple and complex analog ICs. And a hybrid framework, named Image2Net, is proposed for practical conversion from circuit diagrams to netlists. The netlist edit distance (NED) is also introduced to precisely assess the difference between the converted netlists and ground truth. Based on our benchmark, Image2Net achieves 80.77\% successful rate, which is 34.62\%-45.19\% higher than previous works. Specifically, the proposed work shows 0.116 averaged NED, which is 62.1\%-69.6\% lower than state-of-the-arts.
|
[
{
"version": "v1",
"created": "Fri, 27 Jun 2025 07:15:00 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Xu",
"Haohang",
""
],
[
"Liu",
"Chengjie",
""
],
[
"Wang",
"Qihang",
""
],
[
"Huang",
"Wenhao",
""
],
[
"Xu",
"Yongjian",
""
],
[
"Chen",
"Weiyu",
""
],
[
"Peng",
"Anlan",
""
],
[
"Li",
"Zhijun",
""
],
[
"Li",
"Bo",
""
],
[
"Qi",
"Lei",
""
],
[
"Yang",
"Jun",
""
],
[
"Du",
"Yuan",
""
],
[
"Du",
"Li",
""
]
] |
2508.13158
|
Eren Kurshan
|
Yongxiang Liu, Yuchun Ma, Eren Kurshan, Glenn Reinman and Jason Cong
|
Fine Grain 3D Integration for Microarchitecture Design Through Cube Packing Exploration
|
Preprint
|
25th IEEE International Conference on Computer Design, pp. 259-266, 2007
| null | null |
cs.AR cs.ET
|
http://creativecommons.org/licenses/by/4.0/
|
Most previous 3D IC research focused on stacking traditional 2D silicon layers, so the interconnect reduction is limited to inter-block delays. In this paper, we propose techniques that enable efficient exploration of the 3D design space where each logical block can span more than one silicon layers. Although further power and performance improvement is achievable through fine grain 3D integration, the necessary modeling and tool infrastructure has been mostly missing. We develop a cube packing engine which can simultaneously optimize physical and architectural design for effective utilization of 3D in terms of performance, area and temperature. Our experimental results using a design driver show 36% performance improvement (in BIPS) over 2D and 14% over 3D with single layer blocks. Additionally multi-layer blocks can provide up to 30% reduction in power dissipation compared to the single-layer alternatives. Peak temperature of the design is kept within limits as a result of thermal-aware floorplanning and thermal via insertion techniques.
|
[
{
"version": "v1",
"created": "Sun, 13 Jul 2025 15:44:50 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Liu",
"Yongxiang",
""
],
[
"Ma",
"Yuchun",
""
],
[
"Kurshan",
"Eren",
""
],
[
"Reinman",
"Glenn",
""
],
[
"Cong",
"Jason",
""
]
] |
2508.13159
|
Ruibai Tang
|
Ruibai Tang, Wenlai Zhao
|
Accelerating Transistor-Level Simulation of Integrated Circuits via Equivalence of RC Long-Chain Structures
| null | null | null | null |
cs.AR cs.PF
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Transistor-level simulation plays a vital role in validating the physical correctness of integrated circuits. However, such simulations are computationally expensive. This paper proposes three novel reduction methods specifically tailored to RC long-chain structures with different scales of time constant. Such structures account for an average of 6.34\% (up to 12\%) of the total nodes in the benchmark circuits. Experimental results demonstrate that our methods yields an average performance improvement of 8.8\% (up to 22\%) on simulating benchmark circuits which include a variety of functional modules such as ALUs, adders, multipliers, SEC/DED checkers, and interrupt controllers, with only 0.7\% relative error.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 16:43:20 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Tang",
"Ruibai",
""
],
[
"Zhao",
"Wenlai",
""
]
] |
2508.13160
|
Eren Kurshan
|
Yibo Chen, Eren Kurshan, Dave Motschman, Charles Johnson, Yuan Xie
|
Through Silicon Via Aware Design Planning for Thermally Efficient 3-D Integrated Circuits
| null |
IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 1335-1346, 2013
| null | null |
cs.AR cs.ET
|
http://creativecommons.org/licenses/by/4.0/
|
3-D integrated circuits (3-D ICs) offer performance advantages due to their increased bandwidth and reduced wire-length enabled by through-silicon-via structures (TSVs). Traditionally TSVs have been considered to improve the thermal conductivity in the vertical direction. However, the lateral thermal blockage effect becomes increasingly important for TSV via farms (a cluster of TSV vias used for signal bus connections between layers) because the TSV size and pitch continue to scale in {\mu}m range and the metal to insulator ratio becomes smaller. Consequently, dense TSV farms can create lateral thermal blockages in thinned silicon substrate and exacerbate the local hotspots. In this paper, we propose a thermal-aware via farm placement technique for 3-D ICs to minimize lateral heat blockages caused by dense signal bus TSV structures.
|
[
{
"version": "v1",
"created": "Sat, 19 Jul 2025 12:12:06 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Chen",
"Yibo",
""
],
[
"Kurshan",
"Eren",
""
],
[
"Motschman",
"Dave",
""
],
[
"Johnson",
"Charles",
""
],
[
"Xie",
"Yuan",
""
]
] |
2508.13161
|
Zhexuan Xu
|
Zhexuan Xu, Kexin Zhou, Jie Wang, Zijie Geng, Siyuan Xu, Shixiong Kai, Mingxuan Yuan and Feng Wu
|
Piano: A Multi-Constraint Pin Assignment-Aware Floorplanner
| null | null | null | null |
cs.AR cs.AI
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Floorplanning is a critical step in VLSI physical design, increasingly complicated by modern constraints such as fixed-outline requirements, whitespace removal, and the presence of pre-placed modules. In addition, the assignment of pins on module boundaries significantly impacts the performance of subsequent stages, including detailed placement and routing. However, traditional floorplanners often overlook pin assignment with modern constraints during the floorplanning stage. In this work, we introduce Piano, a floorplanning framework that simultaneously optimizes module placement and pin assignment under multiple constraints. Specifically, we construct a graph based on the geometric relationships among modules and their netlist connections, then iteratively search for shortest paths to determine pin assignments. This graph-based method also enables accurate evaluation of feedthrough and unplaced pins, thereby guiding overall layout quality. To further improve the design, we adopt a whitespace removal strategy and employ three local optimizers to enhance layout metrics under multi-constraint scenarios. Experimental results on widely used benchmark circuits demonstrate that Piano achieves an average 6.81% reduction in HPWL, a 13.39% decrease in feedthrough wirelength, a 16.36% reduction in the number of feedthrough modules, and a 21.21% drop in unplaced pins, while maintaining zero whitespace.
|
[
{
"version": "v1",
"created": "Sun, 20 Jul 2025 10:44:54 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Xu",
"Zhexuan",
""
],
[
"Zhou",
"Kexin",
""
],
[
"Wang",
"Jie",
""
],
[
"Geng",
"Zijie",
""
],
[
"Xu",
"Siyuan",
""
],
[
"Kai",
"Shixiong",
""
],
[
"Yuan",
"Mingxuan",
""
],
[
"Wu",
"Feng",
""
]
] |
2508.13162
|
Mahmoud Nazzal
|
Mahmoud Nazzal, Khoa Nguyen, Deepak Vungarala, Ramtin Zand, Shaahin Angizi, Hai Phan, Abdallah Khreishah
|
FedChip: Federated LLM for Artificial Intelligence Accelerator Chip Design
| null | null | null | null |
cs.AR cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
AI hardware design is advancing rapidly, driven by the promise of design automation to make chip development faster, more efficient, and more accessible to a wide range of users. Amongst automation tools, Large Language Models (LLMs) offer a promising solution by automating and streamlining parts of the design process. However, their potential is hindered by data privacy concerns and the lack of domain-specific training. To address this, we introduce FedChip, a Federated fine-tuning approach that enables multiple Chip design parties to collaboratively enhance a shared LLM dedicated for automated hardware design generation while protecting proprietary data. FedChip enables parties to train the model on proprietary local data and improve the shared LLM's performance. To exemplify FedChip's deployment, we create and release APTPU-Gen, a dataset of 30k design variations spanning various performance metric values such as power, performance, and area (PPA). To encourage the LLM to generate designs that achieve a balance across multiple quality metrics, we propose a new design evaluation metric, Chip@k, which statistically evaluates the quality of generated designs against predefined acceptance criteria. Experimental results show that FedChip improves design quality by more than 77% over high-end LLMs while maintaining data privacy
|
[
{
"version": "v1",
"created": "Wed, 23 Jul 2025 15:38:14 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Nazzal",
"Mahmoud",
""
],
[
"Nguyen",
"Khoa",
""
],
[
"Vungarala",
"Deepak",
""
],
[
"Zand",
"Ramtin",
""
],
[
"Angizi",
"Shaahin",
""
],
[
"Phan",
"Hai",
""
],
[
"Khreishah",
"Abdallah",
""
]
] |
2508.13163
|
Yashasvi Makin
|
Yashasvi Makin and Rahul Maliakkal
|
Sustainable AI Training via Hardware-Software Co-Design on NVIDIA, AMD, and Emerging GPU Architectures
|
IEEE CISOSE Industry Track 2025 Conference
| null | null | null |
cs.AR cs.AI cs.DC cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In particular, large-scale deep learning and artificial intelligence model training uses a lot of computational power and energy, so it poses serious sustainability issues. The fast rise in model complexity has resulted in exponential increases in energy consumption, increasing the demand for techniques maximizing computational efficiency and lowering environmental impact. This work explores environmentally driven performance optimization methods especially intended for advanced GPU architectures from NVIDIA, AMD, and other emerging GPU architectures. Our main focus is on investigating hardware-software co-design techniques meant to significantly increase memory-level and kernel-level operations, so improving performance-per-watt measures. Our thorough research encompasses evaluations of specialized tensor and matrix cores, advanced memory optimization methods, and creative integration approaches that taken together result in notable energy efficiency increases. We also discuss important software-level optimizations that augment hardware capability including mixed-precision arithmetic, advanced energy-aware scheduling algorithms, and compiler-driven kernel enhancements. Moreover, we methodically point out important research gaps and suggest future directions necessary to create really sustainable artificial intelligence systems. This paper emphasizes how major increases in training efficiency can be obtained by co-design of hardware and software, so lowering the environmental impact of artificial intelligence without compromising performance. To back up our analysis, we use real-world case studies from top companies like Meta, Google, Amazon, and others that show how these sustainable AI training methods are used in the real world.
|
[
{
"version": "v1",
"created": "Mon, 28 Jul 2025 03:25:44 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Makin",
"Yashasvi",
""
],
[
"Maliakkal",
"Rahul",
""
]
] |
2508.13164
|
Cristina Martin-Do\~nate
|
J. M. Mercado Colmenero and C. Martin Donate
|
A novel geometric predictive algorithm for assessing Compressive Elastic Modulus in MEX additive processes, based on part nonlinearities and layers stiffness,validated with PETG and PLA materials
| null | null |
10.1016/j.polymertesting.2024.108389
| null |
cs.CE
|
http://creativecommons.org/licenses/by/4.0/
|
The paper presents an innovative methodology based on the use of a new predictive algorithm created by the researchers capable of obtaining the elastic modulus of a plastic material manufactured with MEX and its mechanical behaviour in the elastic zone under compressive loads. The predictive algorithm only needs as input the compressive elastic modulus of the isotropic plastic material filament and the manufacturing parameters of the MEX process. The smart developed algorithm calculates the stiffness of each layer considering the number of holes in the projected area. The innovative predictive algorithm has been experimentally and numerically validated using PETG Polyethylene Terephthalate Glycol material and PLA Polylactic Acid on test specimens and on a case study of variable topology. The predictive algorithm is valid for each print pattern and manufacturing direction. The new algorithm improves the existing state of the art significantly since this algorithm extends its utility to most plastic polymer materials suitable for MEX 3D printing, provided that the mechanical and elastic properties of the filament are known. Its versatility extends to complex component geometries subjected to uniaxial compression loads, eliminating the need for mechanical analysis software or expensive experimental validations.
|
[
{
"version": "v1",
"created": "Mon, 28 Jul 2025 11:30:02 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Colmenero",
"J. M. Mercado",
""
],
[
"Donate",
"C. Martin",
""
]
] |
2508.13165
|
Darij Grinberg
|
Darij Grinberg
|
An introduction to the algebra of rings and fields
|
499 pages. Will grow once I teach the course on a semester schedule, but for now it is what it is. Comments are welcome!
| null | null | null |
math.RA math.NT
|
http://creativecommons.org/publicdomain/zero/1.0/
|
This is an introduction to rings and fields, written for a quarter-long undergraduate course. It includes the basic properties of ideals, modules, algebras and polynomials, the constructions of ring extensions and finite fields, some number-theoretical applications (such as a proof of quadratic reciprocity and Jacobsthal's formulas for $p = x^2 + y^2$), and tastes of Gr\"obner bases and the Smith normal form. Familiarity with groups and vector spaces is assumed, though no deep results from either theory are used.
Over 200 exercises are included (mostly without solutions).
|
[
{
"version": "v1",
"created": "Thu, 31 Jul 2025 14:23:28 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Grinberg",
"Darij",
""
]
] |
2508.13167
|
Tianrui Qin
|
Weizhen Li, Jianbo Lin, Zhuosong Jiang, Jingyi Cao, Xinpeng Liu, Jiayu Zhang, Zhenqiang Huang, Qianben Chen, Weichen Sun, Qiexiang Wang, Hongxuan Lu, Tianrui Qin, Chenghao Zhu, Yi Yao, Shuying Fan, Xiaowan Li, Tiannan Wang, Pai Liu, King Zhu, He Zhu, Dingfeng Shi, Piaohong Wang, Yeyi Guan, Xiangru Tang, Minghao Liu, Yuchen Eleanor Jiang, Jian Yang, Jiaheng Liu, Ge Zhang, Wangchunshu Zhou
|
Chain-of-Agents: End-to-End Agent Foundation Models via Multi-Agent Distillation and Agentic RL
|
51 pages
| null | null | null |
cs.AI cs.CL
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Recent advances in large language models (LLMs) and multi-agent systems have demonstrated remarkable capabilities in complex problem-solving tasks such as deep research, vibe coding, and mathematical reasoning. However, most existing multi-agent systems are built upon manual prompt/workflow engineering with sophisticated agent frameworks, making them computationally inefficient, less capable, and can not benefit from data-centric learning. In this work, we introduce Chain-of-Agents (CoA), a novel paradigm of LLM reasoning that enables native end-to-end complex problem-solving in the same way as a multi-agent system (i.e., multi-turn problem solving with multiple tools and multiple agents) within one model. In chain-of-agents problem-solving, the model dynamically activates different tool agents and role-playing agents to simulate multi-agent collaboration in an end-to-end fashion. To elicit end-to-end chain-of-agents problem-solving abilities in LLMs, we introduce a multi-agent distillation framework to distill state-of-the-art multi-agent systems into chain-of-agents trajectories for agentic supervised fine-tuning. We then use agentic reinforcement learning on verifiable agentic tasks to further improve the models' capabilities on chain-of-agents problem solving. We call the resulting models Agent Foundation Models (AFMs). Our empirical studies demonstrate that AFM establishes new state-of-the-art performance across diverse benchmarks in both web agent and code agent settings. We make the entire research, including the model weights, code for training and evaluation, and the training data, fully open-sourced, which offers a solid starting point for future research on agent models and agentic RL.
|
[
{
"version": "v1",
"created": "Wed, 06 Aug 2025 17:01:02 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Li",
"Weizhen",
""
],
[
"Lin",
"Jianbo",
""
],
[
"Jiang",
"Zhuosong",
""
],
[
"Cao",
"Jingyi",
""
],
[
"Liu",
"Xinpeng",
""
],
[
"Zhang",
"Jiayu",
""
],
[
"Huang",
"Zhenqiang",
""
],
[
"Chen",
"Qianben",
""
],
[
"Sun",
"Weichen",
""
],
[
"Wang",
"Qiexiang",
""
],
[
"Lu",
"Hongxuan",
""
],
[
"Qin",
"Tianrui",
""
],
[
"Zhu",
"Chenghao",
""
],
[
"Yao",
"Yi",
""
],
[
"Fan",
"Shuying",
""
],
[
"Li",
"Xiaowan",
""
],
[
"Wang",
"Tiannan",
""
],
[
"Liu",
"Pai",
""
],
[
"Zhu",
"King",
""
],
[
"Zhu",
"He",
""
],
[
"Shi",
"Dingfeng",
""
],
[
"Wang",
"Piaohong",
""
],
[
"Guan",
"Yeyi",
""
],
[
"Tang",
"Xiangru",
""
],
[
"Liu",
"Minghao",
""
],
[
"Jiang",
"Yuchen Eleanor",
""
],
[
"Yang",
"Jian",
""
],
[
"Liu",
"Jiaheng",
""
],
[
"Zhang",
"Ge",
""
],
[
"Zhou",
"Wangchunshu",
""
]
] |
2508.13168
|
Franco Bagnoli
|
Franco Bagnoli and Tommaso Matteuzzi
|
Metastability in the parallel Ising model
| null | null | null | null |
cond-mat.stat-mech
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this short paper we present some considerations about some parallel implementations of the dynamic (Monte Carlo) version of the Ising model. In some cases the equilibrium distribution of the parallel version does not present the symmetry breaking phenomenon in the low-temperature phase, i.e., the stochastic trajectory originated by the Monte Carlo simulation can jump between the distributions corresponding to both kinds of magnetization, or the lattice can break into two disjoint sublattices, each of which goes into a different asymptotic distribution. In this latter case, by introducing a small asynchronism, we can have a transition towards the standard ferromagnetic (or antiferromagnetic) distribution, with metastable transients.
|
[
{
"version": "v1",
"created": "Thu, 07 Aug 2025 06:03:36 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Bagnoli",
"Franco",
""
],
[
"Matteuzzi",
"Tommaso",
""
]
] |
2508.13169
|
Stefanie Urchs
|
Stefanie Urchs, Veronika Thurner, Matthias A{\ss}enmacher, Christian Heumann, Stephanie Thiemichen
|
Fair Play in the Newsroom: Actor-Based Filtering Gender Discrimination in Text Corpora
| null | null | null | null |
cs.CL cs.CY
|
http://creativecommons.org/licenses/by/4.0/
|
Large language models are increasingly shaping digital communication, yet their outputs often reflect structural gender imbalances that originate from their training data. This paper presents an extended actor-level pipeline for detecting and mitigating gender discrimination in large-scale text corpora. Building on prior work in discourse-aware fairness analysis, we introduce new actor-level metrics that capture asymmetries in sentiment, syntactic agency, and quotation styles. The pipeline supports both diagnostic corpus analysis and exclusion-based balancing, enabling the construction of fairer corpora. We apply our approach to the taz2024full corpus of German newspaper articles from 1980 to 2024, demonstrating substantial improvements in gender balance across multiple linguistic dimensions. Our results show that while surface-level asymmetries can be mitigated through filtering and rebalancing, subtler forms of bias persist, particularly in sentiment and framing. We release the tools and reports to support further research in discourse-based fairness auditing and equitable corpus construction.
|
[
{
"version": "v1",
"created": "Thu, 07 Aug 2025 11:43:58 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Urchs",
"Stefanie",
""
],
[
"Thurner",
"Veronika",
""
],
[
"Aßenmacher",
"Matthias",
""
],
[
"Heumann",
"Christian",
""
],
[
"Thiemichen",
"Stephanie",
""
]
] |
2508.13170
|
Ugur Tirnakli
|
Abbas Ali Saberi, Ugur Tirnakli and Constantino Tsallis
|
Central Limit Behavior at the Edge of Chaos in the z-Logistic Map
|
15 pages, 5 figures
| null | null | null |
cond-mat.stat-mech nlin.CD
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We focus on the FeigenbaumCoulletTresser point of the dissipative one-dimensional z logistic map. We show that sums of iterates converge to q Gaussian distributions, which optimize the nonadditive entropic functional Sq under simple constraints. We derive a closedform prediction for the entropic index, and validate it numerically via data collapse for typical z values. The formula captures how the limiting law depends on the nonlinearity order and implies finite variance for z larger than 2 and divergent variance for z in between 1 and 2. These results extend edge of chaos central limit behavior beyond the standard case and provide a simple predictive law for unimodal maps with varying maximum order.
|
[
{
"version": "v1",
"created": "Fri, 08 Aug 2025 08:06:58 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Saberi",
"Abbas Ali",
""
],
[
"Tirnakli",
"Ugur",
""
],
[
"Tsallis",
"Constantino",
""
]
] |
2508.13171
|
Tao An
|
Tao An
|
Cognitive Workspace: Active Memory Management for LLMs -- An Empirical Study of Functional Infinite Context
|
13 pages, 1 figure, code available at https://github.com/tao-hpu/cognitive-workspace
| null | null | null |
cs.AI cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
Large Language Models (LLMs) face fundamental limitations in context management despite recent advances extending context windows to millions of tokens. We propose Cognitive Workspace, a novel paradigm that transcends traditional Retrieval-Augmented Generation (RAG) by emulating human cognitive mechanisms of external memory use. Drawing from cognitive science foundations including Baddeley's working memory model, Clark's extended mind thesis, and Hutchins' distributed cognition framework, we demonstrate that current passive retrieval systems fail to capture the dynamic, task-driven nature of human memory management. Our analysis of 2024-2025 developments reveals that while techniques like Infini-attention and StreamingLLM achieve impressive context lengths, they lack the metacognitive awareness and active planning capabilities essential for true cognitive extension. Cognitive Workspace addresses these limitations through three core innovations: (1) active memory management with deliberate information curation, (2) hierarchical cognitive buffers enabling persistent working states, and (3) task-driven context optimization that dynamically adapts to cognitive demands. Empirical validation demonstrates Cognitive Workspace achieves an average 58.6% memory reuse rate (ranging from 54-60% across different tasks) compared to 0% for traditional RAG, with 17-18% net efficiency gain despite 3.3x higher operation counts. Statistical analysis confirms these advantages with p < 0.001 and Cohen's d > 23 across multiple task types, establishing the first quantitative evidence for active memory superiority in LLM systems. We present a comprehensive theoretical framework synthesizing insights from 50+ recent papers, positioning Cognitive Workspace as a fundamental shift from information retrieval to genuine cognitive augmentation.
|
[
{
"version": "v1",
"created": "Fri, 08 Aug 2025 16:32:47 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"An",
"Tao",
""
]
] |
2508.13172
|
Jianqiu Chen
|
Jianqiu Chen, Siqi Li, Xu He
|
White-Box Reasoning: Synergizing LLM Strategy and gm/Id Data for Automated Analog Circuit Design
|
8 pages, 4 figures, 7 Tables
| null | null | null |
cs.AR cs.AI cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Analog IC design is a bottleneck due to its reliance on experience and inefficient simulations, as traditional formulas fail in advanced nodes. Applying Large Language Models (LLMs) directly to this problem risks mere "guessing" without engineering principles. We present a "synergistic reasoning" framework that integrates an LLM's strategic reasoning with the physical precision of the gm/Id methodology. By empowering the LLM with gm/Id lookup tables, it becomes a quantitative, data-driven design partner.
We validated this on a two-stage op-amp, where our framework enabled the Gemini model to meet all TT corner specs in 5 iterations and extended optimization to all PVT corners. A crucial ablation study proved gm/Id data is key for this efficiency and precision; without it, the LLM is slower and deviates. Compared to a senior engineer's design, our framework achieves quasi-expert quality with an order-of-magnitude improvement in efficiency. This work validates a path for true analog design automation by combining LLM reasoning with scientific circuit design methodologies.
|
[
{
"version": "v1",
"created": "Sat, 09 Aug 2025 01:25:27 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Chen",
"Jianqiu",
""
],
[
"Li",
"Siqi",
""
],
[
"He",
"Xu",
""
]
] |
2508.13174
|
Binqi Chen
|
Hongjun Ding, Binqi Chen, Jinsheng Huang, Taian Guo, Zhengyang Mao, Guoyi Shao, Lutong Zou, Luchen Liu, Ming Zhang
|
AlphaEval: A Comprehensive and Efficient Evaluation Framework for Formula Alpha Mining
|
12 pages, 5 figures
| null | null | null |
cs.AI cs.LG q-fin.CP stat.ML
|
http://creativecommons.org/licenses/by/4.0/
|
Formula alpha mining, which generates predictive signals from financial data, is critical for quantitative investment. Although various algorithmic approaches-such as genetic programming, reinforcement learning, and large language models-have significantly expanded the capacity for alpha discovery, systematic evaluation remains a key challenge. Existing evaluation metrics predominantly include backtesting and correlation-based measures. Backtesting is computationally intensive, inherently sequential, and sensitive to specific strategy parameters. Correlation-based metrics, though efficient, assess only predictive ability and overlook other crucial properties such as temporal stability, robustness, diversity, and interpretability. Additionally, the closed-source nature of most existing alpha mining models hinders reproducibility and slows progress in this field. To address these issues, we propose AlphaEval, a unified, parallelizable, and backtest-free evaluation framework for automated alpha mining models. AlphaEval assesses the overall quality of generated alphas along five complementary dimensions: predictive power, stability, robustness to market perturbations, financial logic, and diversity. Extensive experiments across representative alpha mining algorithms demonstrate that AlphaEval achieves evaluation consistency comparable to comprehensive backtesting, while providing more comprehensive insights and higher efficiency. Furthermore, AlphaEval effectively identifies superior alphas compared to traditional single-metric screening approaches. All implementations and evaluation tools are open-sourced to promote reproducibility and community engagement.
|
[
{
"version": "v1",
"created": "Sun, 10 Aug 2025 11:19:24 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Ding",
"Hongjun",
""
],
[
"Chen",
"Binqi",
""
],
[
"Huang",
"Jinsheng",
""
],
[
"Guo",
"Taian",
""
],
[
"Mao",
"Zhengyang",
""
],
[
"Shao",
"Guoyi",
""
],
[
"Zou",
"Lutong",
""
],
[
"Liu",
"Luchen",
""
],
[
"Zhang",
"Ming",
""
]
] |
2508.13175
|
Xavier Urbain
|
Pierre Guichard, Arnaud Dochain, Rapha\"el Marion, Pauline de Crombrugghe de Picquendaele, Nicolas Lejeune, Beno\^it Hackens, Paul-Antoine Hervieux, and Xavier Urbain
|
Fast hydrogen atom diffraction through monocrystalline graphene
|
6 pages and 5 figures (main text), 6 pages and 5 figures (supplemental material)
| null | null | null |
cond-mat.mes-hall physics.atom-ph quant-ph
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
We report fast atom diffraction through single-layer graphene using hydrogen atoms at kinetic energies from 150 to 1200 eV. High-resolution images reveal overlapping hexagonal patterns from coexisting monocrystalline domains. Time-of-flight tagging confirms negligible energy loss, making the method suitable for matter-wave interferometry. The diffraction is well described by the eikonal approximation, with accurate modeling requiring the full 3D interaction potential from DFT. Simpler models fail to reproduce the data, highlighting the exceptional sensitivity of diffraction patterns to atom-surface interactions and their potential for spectroscopic applications.
|
[
{
"version": "v1",
"created": "Mon, 11 Aug 2025 08:02:11 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Guichard",
"Pierre",
""
],
[
"Dochain",
"Arnaud",
""
],
[
"Marion",
"Raphaël",
""
],
[
"de Picquendaele",
"Pauline de Crombrugghe",
""
],
[
"Lejeune",
"Nicolas",
""
],
[
"Hackens",
"Benoît",
""
],
[
"Hervieux",
"Paul-Antoine",
""
],
[
"Urbain",
"Xavier",
""
]
] |
2508.13176
|
Simon Hosemann
|
Simon Hosemann, Jean Christoph Jung, Carsten Lutz, Sebastian Rudolph
|
Fitting Ontologies and Constraints to Relational Structures
|
Accepted at the 22nd International Conference on Principles of Knowledge Representation and Reasoning (KR 2025)
| null | null | null |
cs.AI cs.DB
|
http://creativecommons.org/licenses/by/4.0/
|
We study the problem of fitting ontologies and constraints to positive and negative examples that take the form of a finite relational structure. As ontology and constraint languages, we consider the description logics $\mathcal{E\mkern-2mu L}$ and $\mathcal{E\mkern-2mu LI}$ as well as several classes of tuple-generating dependencies (TGDs): full, guarded, frontier-guarded, frontier-one, and unrestricted TGDs as well as inclusion dependencies. We pinpoint the exact computational complexity, design algorithms, and analyze the size of fitting ontologies and TGDs. We also investigate the related problem of constructing a finite basis of concept inclusions / TGDs for a given set of finite structures. While finite bases exist for $\mathcal{E\mkern-2mu L}$, $\mathcal{E\mkern-2mu LI}$, guarded TGDs, and inclusion dependencies, they in general do not exist for full, frontier-guarded and frontier-one TGDs.
|
[
{
"version": "v1",
"created": "Mon, 11 Aug 2025 21:52:58 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Hosemann",
"Simon",
""
],
[
"Jung",
"Jean Christoph",
""
],
[
"Lutz",
"Carsten",
""
],
[
"Rudolph",
"Sebastian",
""
]
] |
2508.13177
|
Miguel De Prado
|
Nikola Pi\v{z}urica, Nikola Milovi\'c, Igor Jovan\v{c}evi\'c, Conor Heins, and Miguel de Prado
|
A Hardware-oriented Approach for Efficient Active Inference Computation and Deployment
| null | null | null | null |
cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Active Inference (AIF) offers a robust framework for decision-making, yet its computational and memory demands pose challenges for deployment, especially in resource-constrained environments. This work presents a methodology that facilitates AIF's deployment by integrating pymdp's flexibility and efficiency with a unified, sparse, computational graph tailored for hardware-efficient execution. Our approach reduces latency by over 2x and memory by up to 35%, advancing the deployment of efficient AIF agents for real-time and embedded applications.
|
[
{
"version": "v1",
"created": "Tue, 12 Aug 2025 09:39:46 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Pižurica",
"Nikola",
""
],
[
"Milović",
"Nikola",
""
],
[
"Jovančević",
"Igor",
""
],
[
"Heins",
"Conor",
""
],
[
"de Prado",
"Miguel",
""
]
] |
2508.13178
|
Cong Zhang
|
Cong Zhang
|
The Interpretability Analysis of the Model Can Bring Improvements to the Text-to-SQL Task
| null | null | null | null |
cs.AI cs.CL cs.DB
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
To elevate the foundational capabilities and generalization prowess of the text-to-SQL model in real-world applications, we integrate model interpretability analysis with execution-guided strategy for semantic parsing of WHERE clauses in SQL queries. Furthermore, we augment this approach with filtering adjustments, logical correlation refinements, and model fusion, culminating in the design of the CESQL model that facilitates conditional enhancement. Our model excels on the WikiSQL dataset, which is emblematic of single-table database query tasks, markedly boosting the accuracy of prediction outcomes. When predicting conditional values in WHERE clauses, we have not only minimized our dependence on data within the condition columns of tables but also circumvented the impact of manually labeled training data. Our hope is that this endeavor to enhance accuracy in processing basic database queries will offer fresh perspectives for research into handling complex queries and scenarios featuring irregular data in real-world database environments.
|
[
{
"version": "v1",
"created": "Tue, 12 Aug 2025 11:24:16 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Zhang",
"Cong",
""
]
] |
2508.13179
|
Samuel Segun
|
Samuel T. Segun, Rachel Adams, Ana Florido, Scott Timcke, Jonathan Shock, Leah Junck, Fola Adeleke, Nicolas Grossman, Ayantola Alayande, Jerry John Kponyo, Matthew Smith, Dickson Marfo Fosu, Prince Dawson Tetteh, Juliet Arthur, Stephanie Kasaon, Odilile Ayodele, Laetitia Badolo, Paul Plantinga, Michael Gastrow, Sumaya Nur Adan, Joanna Wiaterek, Cecil Abungu, Kojo Apeagyei, Luise Eder, Tegawende Bissyande
|
Toward an African Agenda for AI Safety
|
28 pages, 2 figures
| null | null | null |
cs.CY cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
This paper maps Africa's distinctive AI risk profile, from deepfake fuelled electoral interference and data colonial dependency to compute scarcity, labour disruption and disproportionate exposure to climate driven environmental costs. While major benefits are promised to accrue, the availability, development and adoption of AI also mean that African people and countries face particular AI safety risks, from large scale labour market disruptions to the nefarious use of AI to manipulate public opinion. To date, African perspectives have not been meaningfully integrated into global debates and processes regarding AI safety, leaving African stakeholders with limited influence over the emerging global AI safety governance agenda. While there are Computer Incident Response Teams on the continent, none hosts a dedicated AI Safety Institute or office. We propose a five-point action plan centred on (i) a policy approach that foregrounds the protection of the human rights of those most vulnerable to experiencing the harmful socio-economic effects of AI; (ii) the establishment of an African AI Safety Institute; (iii) promote public AI literacy and awareness; (iv) development of early warning system with inclusive benchmark suites for 25+ African languages; and (v) an annual AU-level AI Safety & Security Forum.
|
[
{
"version": "v1",
"created": "Tue, 12 Aug 2025 17:42:09 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Segun",
"Samuel T.",
""
],
[
"Adams",
"Rachel",
""
],
[
"Florido",
"Ana",
""
],
[
"Timcke",
"Scott",
""
],
[
"Shock",
"Jonathan",
""
],
[
"Junck",
"Leah",
""
],
[
"Adeleke",
"Fola",
""
],
[
"Grossman",
"Nicolas",
""
],
[
"Alayande",
"Ayantola",
""
],
[
"Kponyo",
"Jerry John",
""
],
[
"Smith",
"Matthew",
""
],
[
"Fosu",
"Dickson Marfo",
""
],
[
"Tetteh",
"Prince Dawson",
""
],
[
"Arthur",
"Juliet",
""
],
[
"Kasaon",
"Stephanie",
""
],
[
"Ayodele",
"Odilile",
""
],
[
"Badolo",
"Laetitia",
""
],
[
"Plantinga",
"Paul",
""
],
[
"Gastrow",
"Michael",
""
],
[
"Adan",
"Sumaya Nur",
""
],
[
"Wiaterek",
"Joanna",
""
],
[
"Abungu",
"Cecil",
""
],
[
"Apeagyei",
"Kojo",
""
],
[
"Eder",
"Luise",
""
],
[
"Bissyande",
"Tegawende",
""
]
] |
2508.13180
|
Zifan Wang
|
Ziwen Han, Meher Mankikar, Julian Michael, Zifan Wang
|
Search-Time Data Contamination
| null | null | null | null |
cs.AI cs.LG
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Data contamination refers to the leakage of evaluation data into model training data, resulting in overfitting to supposedly held-out test sets and compromising test validity. We identify an analogous issue, search-time contamination (STC), in evaluating search-based LLM agents which use tools to gather information from online sources when answering user queries. STC occurs when the retrieval step surfaces a source containing the test question (or a near-duplicate) alongside its answer, enabling agents to copy rather than genuinely infer or reason, undermining benchmark integrity. We find that HuggingFace, an online platform hosting evaluation datasets, appears among retrieved sources in search based agent logs. Consequently, agents often explicitly acknowledge discovering question answer pairs from HuggingFace within their reasoning chains. On three commonly used capability benchmarks: Humanity's Last Exam (HLE), SimpleQA, and GPQA, we demonstrate that for approximately 3% of questions, search-based agents directly find the datasets with ground truth labels on HuggingFace. When millions of evaluation queries target the same benchmark, even small, repeated leaks can accelerate the benchmark's obsolescence, shortening its intended lifecycle. After HuggingFace is blocked, we observe a drop in accuracy on the contaminated subset of approximately 15%. We further show through ablation experiments that publicly accessible evaluation datasets on HuggingFace may not be the sole source of STC. To this end, we conclude by proposing best practices for benchmark design and result reporting to address this novel form of leakage and ensure trustworthy evaluation of search-based LLM agents. To facilitate the auditing of evaluation results, we also publicly release the complete logs from our experiments.
|
[
{
"version": "v1",
"created": "Tue, 12 Aug 2025 22:52:21 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Han",
"Ziwen",
""
],
[
"Mankikar",
"Meher",
""
],
[
"Michael",
"Julian",
""
],
[
"Wang",
"Zifan",
""
]
] |
2508.13181
|
Dominik Marek Loroch
|
Dominik Loroch, Johannes Feldmann, Vladimir Rybalkin, Norbert Wehn
|
Low-power, Energy-efficient, Cardiologist-level Atrial Fibrillation Detection for Wearable Devices
|
2025 IEEE 38th International System-on-Chip Conference (SOCC)
| null | null | null |
cs.AR eess.SP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Atrial fibrillation (AF) is a common arrhythmia and major risk factor for cardiovascular complications. While commercially available devices and supporting Artificial Intelligence (AI) algorithms exist for reliable detection of AF, the scaling of this technology to the amount of people who need this diagnosis is still a major challenge. This paper presents a novel wearable device, designed specifically for the early and reliable detection of AF. We present an FPGA-based patch-style wearable monitor with embedded deep learning-based AF detection. Operating with 3.8mW system power, which is 1-3 orders of magnitude lower than the state-of-the-art, the device enables continuous AF detection for over three weeks while achieving 95% accuracy, surpassing cardiologist-level performance. A key innovation is the combination of energy-efficient hardware-software co-design and optimized power management through the application of hardware-aware neural architecture search. This advancement represents a significant step toward scalable, reliable, and sustainable AF monitoring.
|
[
{
"version": "v1",
"created": "Wed, 13 Aug 2025 12:11:54 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Loroch",
"Dominik",
""
],
[
"Feldmann",
"Johannes",
""
],
[
"Rybalkin",
"Vladimir",
""
],
[
"Wehn",
"Norbert",
""
]
] |
2508.13182
|
Prateek Ranka
|
Prateek Ranka, Fred Morstatter, Andrea Belz and Alexandra Graddy-Reed
|
Using Artificial Intuition in Distinct, Minimalist Classification of Scientific Abstracts for Management of Technology Portfolios
| null | null | null | null |
cs.DL cs.AI cs.LG
|
http://creativecommons.org/publicdomain/zero/1.0/
|
Classification of scientific abstracts is useful for strategic activities but challenging to automate because the sparse text provides few contextual clues. Metadata associated with the scientific publication can be used to improve performance but still often requires a semi-supervised setting. Moreover, such schemes may generate labels that lack distinction -- namely, they overlap and thus do not uniquely define the abstract. In contrast, experts label and sort these texts with ease. Here we describe an application of a process we call artificial intuition to replicate the expert's approach, using a Large Language Model (LLM) to generate metadata. We use publicly available abstracts from the United States National Science Foundation to create a set of labels, and then we test this on a set of abstracts from the Chinese National Natural Science Foundation to examine funding trends. We demonstrate the feasibility of this method for research portfolio management, technology scouting, and other strategic activities.
|
[
{
"version": "v1",
"created": "Wed, 13 Aug 2025 22:32:39 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Ranka",
"Prateek",
""
],
[
"Morstatter",
"Fred",
""
],
[
"Belz",
"Andrea",
""
],
[
"Graddy-Reed",
"Alexandra",
""
]
] |
2508.13184
|
Tai Vu
|
Tai Vu, Robert Yang
|
BERT-VQA: Visual Question Answering on Plots
| null | null | null | null |
cs.LG cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Visual question answering has been an exciting challenge in the field of natural language understanding, as it requires deep learning models to exchange information from both vision and language domains. In this project, we aim to tackle a subtask of this problem, namely visual question answering on plots. To achieve this, we developed BERT-VQA, a VisualBERT-based model architecture with a pretrained ResNet 101 image encoder, along with a potential addition of joint fusion. We trained and evaluated this model against a baseline that consisted of a LSTM, a CNN, and a shallow classifier. The final outcome disproved our core hypothesis that the cross-modality module in VisualBERT is essential in aligning plot components with question phrases. Therefore, our work provided valuable insights into the difficulty of the plot question answering challenge as well as the appropriateness of different model architectures in solving this problem.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 00:55:18 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Vu",
"Tai",
""
],
[
"Yang",
"Robert",
""
]
] |
2508.13185
|
Jun Cui
|
Jun Cui
|
Digital-GenAI-Enhanced HCI in DevOps as a Driver of Sustainable Innovation: An Empirical Framework
| null | null | null | null |
cs.CY
|
http://creativecommons.org/publicdomain/zero/1.0/
|
This study examines the impact of Digital-GenAI-Enhanced Human-Computer Interaction (HCI) in DevOps on sustainable innovation performance among Chinese A-share internet technology firms. Using panel data from 2018-2024, we analyze 5,560 firm-year observations from CNRDS and CSMAR databases. Our empirical framework reveals significant positive associations between AI-enhanced HCI implementation and sustainable innovation outcomes. Results demonstrate that firms adopting advanced HCI technologies achieve 23.7% higher innovation efficiency. The study contributes to understanding digital transformation's role in sustainable business practices. We identify three key mechanisms: operational efficiency enhancement, knowledge integration facilitation, and stakeholder engagement improvement. Findings provide practical implications for technology adoption strategies in emerging markets
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 13:05:34 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Cui",
"Jun",
""
]
] |
2508.13186
|
Xingyuan Bu
|
Shilong Li, Xingyuan Bu, Wenjie Wang, Jiaheng Liu, Jun Dong, Haoyang He, Hao Lu, Haozhe Zhang, Chenchen Jing, Zhen Li, Chuanhao Li, Jiayi Tian, Chenchen Zhang, Tianhao Peng, Yancheng He, Jihao Gu, Yuanxing Zhang, Jian Yang, Ge Zhang, Wenhao Huang, Wangchunshu Zhou, Zhaoxiang Zhang, Ruizhe Ding, Shilei Wen
|
MM-BrowseComp: A Comprehensive Benchmark for Multimodal Browsing Agents
|
The first two authors contribute equally, 26 pages, repo at https://github.com/MMBrowseComp/MM-BrowseComp
| null | null | null |
cs.CL cs.AI cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
AI agents with advanced reasoning and tool use capabilities have demonstrated impressive performance in web browsing for deep search. While existing benchmarks such as BrowseComp evaluate these browsing abilities, they primarily focus on textual information, overlooking the prevalence of multimodal content. To bridge this gap, we introduce MM-BrowseComp, a novel benchmark comprising 224 challenging, hand-crafted questions specifically designed to assess agents' multimodal retrieval and reasoning capabilities. These questions often incorporate images in prompts, and crucial information encountered during the search and reasoning process may also be embedded within images or videos on webpages. Consequently, methods relying solely on text prove insufficient for our benchmark. Additionally, we provide a verified checklist for each question, enabling fine-grained analysis of multimodal dependencies and reasoning paths. Our comprehensive evaluation of state-of-the-art models on MM-BrowseComp reveals that even top models like OpenAI o3 with tools achieve only 29.02\% accuracy, highlighting the suboptimal multimodal capabilities and lack of native multimodal reasoning in current models.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 13:46:47 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Li",
"Shilong",
""
],
[
"Bu",
"Xingyuan",
""
],
[
"Wang",
"Wenjie",
""
],
[
"Liu",
"Jiaheng",
""
],
[
"Dong",
"Jun",
""
],
[
"He",
"Haoyang",
""
],
[
"Lu",
"Hao",
""
],
[
"Zhang",
"Haozhe",
""
],
[
"Jing",
"Chenchen",
""
],
[
"Li",
"Zhen",
""
],
[
"Li",
"Chuanhao",
""
],
[
"Tian",
"Jiayi",
""
],
[
"Zhang",
"Chenchen",
""
],
[
"Peng",
"Tianhao",
""
],
[
"He",
"Yancheng",
""
],
[
"Gu",
"Jihao",
""
],
[
"Zhang",
"Yuanxing",
""
],
[
"Yang",
"Jian",
""
],
[
"Zhang",
"Ge",
""
],
[
"Huang",
"Wenhao",
""
],
[
"Zhou",
"Wangchunshu",
""
],
[
"Zhang",
"Zhaoxiang",
""
],
[
"Ding",
"Ruizhe",
""
],
[
"Wen",
"Shilei",
""
]
] |
2508.13187
|
Jonathan Karr Jr
|
Jonathan A. Karr Jr., Benjamin F. Herbst, Ting Hua, Matthew Hauenstein, Georgina Curto, Nitesh V. Chawla
|
Combating Homelessness Stigma with LLMs: A New Multi-Modal Dataset for Bias Detection
| null | null | null | null |
cs.CY cs.AI cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
Homelessness is a persistent social challenge, impacting millions worldwide. Over 770,000 people experienced homelessness in the U.S. in 2024. Social stigmatization is a significant barrier to alleviation, shifting public perception, and influencing policymaking. Given that online and city council discourse reflect and influence part of public opinion, it provides valuable insights to identify and track social biases. This research contributes to alleviating homelessness by acting on public opinion. It introduces novel methods, building on natural language processing (NLP) and large language models (LLMs), to identify and measure PEH social bias expressed in digital spaces. We present a new, manually-annotated multi-modal dataset compiled from Reddit, X (formerly Twitter), news articles, and city council meeting minutes across 10 U.S. cities. This unique dataset provides evidence of the typologies of homelessness bias described in the literature. In order to scale up and automate the detection of homelessness bias online, we evaluate LLMs as classifiers. We applied both zero-shot and few-shot classification techniques to this data. We utilized local LLMs (Llama 3.2 3B Instruct, Qwen 2.5 7B Instruct, and Phi4 Instruct Mini) as well as closed-source API models (GPT-4.1, Gemini 2.5 Pro, and Grok-4). Our findings reveal that although there are significant inconsistencies in local LLM zero-shot classification, the in-context learning classification scores of local LLMs approach the classification scores of closed-source LLMs. Furthermore, LLMs outperform BERT when averaging across all categories. This work aims to raise awareness about the pervasive bias against PEH, develop new indicators to inform policy, and ultimately enhance the fairness and ethical application of Generative AI technologies.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 17:58:34 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Karr",
"Jonathan A.",
"Jr."
],
[
"Herbst",
"Benjamin F.",
""
],
[
"Hua",
"Ting",
""
],
[
"Hauenstein",
"Matthew",
""
],
[
"Curto",
"Georgina",
""
],
[
"Chawla",
"Nitesh V.",
""
]
] |
2508.13188
|
Md Al Amin
|
Md Al Amin, Bikash Kumar Paul
|
Colon Polyps Detection from Colonoscopy Images Using Deep Learning
|
17 Pages
| null | null | null |
eess.IV cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Colon polyps are precursors to colorectal cancer, a leading cause of cancer-related mortality worldwide. Early detection is critical for improving patient outcomes. This study investigates the application of deep learning-based object detection for early polyp identification using colonoscopy images. We utilize the Kvasir-SEG dataset, applying extensive data augmentation and splitting the data into training (80\%), validation (20\% of training), and testing (20\%) sets. Three variants of the YOLOv5 architecture (YOLOv5s, YOLOv5m, YOLOv5l) are evaluated. Experimental results show that YOLOv5l outperforms the other variants, achieving a mean average precision (mAP) of 85.1\%, with the highest average Intersection over Union (IoU) of 0.86. These findings demonstrate that YOLOv5l provides superior detection performance for colon polyp localization, offering a promising tool for enhancing colorectal cancer screening accuracy.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 20:16:23 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Amin",
"Md Al",
""
],
[
"Paul",
"Bikash Kumar",
""
]
] |
2508.13189
|
Chirag Nagpal
|
Chirag Nagpal
|
Preference Models assume Proportional Hazards of Utilities
| null | null | null | null |
stat.ML cs.AI cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Approaches for estimating preferences from human annotated data typically involves inducing a distribution over a ranked list of choices such as the Plackett-Luce model. Indeed, modern AI alignment tools such as Reward Modelling and Direct Preference Optimization are based on the statistical assumptions posed by the Plackett-Luce model. In this paper, I will connect the Plackett-Luce model to another classical and well known statistical model, the Cox Proportional Hazards model and attempt to shed some light on the implications of the connection therein.
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 00:08:56 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Nagpal",
"Chirag",
""
]
] |
2508.13190
|
Jinrui Zhou
|
Jinrui Zhou, Yiqi Gu, Song Jiang, Hua Shen, Liwei Xu, Guanyu Zhou
|
Data-driven optimized high-order WENO schemes with low-dissipation and low-dispersion
|
22 pages, 34 figures
| null | null | null |
math.NA cs.NA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Classical high-order weighted essentially non-oscillatory (WENO) schemes are designed to achieve optimal convergence order for smooth solutions and to maintain non-oscillatory behaviors for discontinuities. However, their spectral properties are not optimal, which limits the ability to capture high-frequency waves and small-scale features. In this paper, we propose a data-driven optimized method to improve the spectral properties of the WENO schemes. By analyzing the approximate dispersion relation (ADR), the spectral error of the schemes can be bounded by the reconstructed errors of a series of trigonometric functions with different wavenumbers. Therefore, we propose the new schemes WENO5-JS/Z-NN that introduce a compensation term parameterized by a neural network to the weight function of the WENO5-JS/Z schemes. The neural network is trained such that the generated weights can minimize the reconstructed errors over a large number of spatial stencils, and furthermore, improve the spectral accuracy. Meanwhile, the Total Variation Diminishing (TVD) constraint and anti-dissipation penalization are incorporated into the loss function to enhance the shock-capturing capability and preserve stability in simulating high-frequency waves. Compared to WENO5-JS/Z, our schemes maintain the ability to capture discontinuities while providing higher resolution for fine-scale flow features. The ADR indicates that the new schemes can match the exact spectrum more accurately over a broader range of wavenumbers.
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 05:29:29 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Zhou",
"Jinrui",
""
],
[
"Gu",
"Yiqi",
""
],
[
"Jiang",
"Song",
""
],
[
"Shen",
"Hua",
""
],
[
"Xu",
"Liwei",
""
],
[
"Zhou",
"Guanyu",
""
]
] |
2508.13191
|
Ke Ding
|
Ke Ding, Brian Parker, Jiayu Wen
|
NucEL: Single-Nucleotide ELECTRA-Style Genomic Pre-training for Efficient and Interpretable Representations
| null | null | null | null |
q-bio.GN
|
http://creativecommons.org/licenses/by/4.0/
|
Pre-training large language models on genomic sequences is a powerful approach for learning biologically meaningful representations. Masked language modeling (MLM) methods, such as DNABERT and Nucleotide Transformer (NT), achieve strong performance but suffer from partial token supervision, pre-training/fine-tuning mismatches, and high computational costs. We introduce NucEL, the first ELECTRA-style pre-training framework for genomic foundation models, addressing these limitations. Using a discriminator to identify tokens altered by a generator, NucEL provides comprehensive token-level supervision across all sequence positions, improving efficiency over the partial supervision of MLM. Incorporating ModernBERT's hybrid local-global attention and flash attention, NucEL offers an optimized BERT architecture for genomic modeling. Unlike 6-mer tokenization, NucEL uses single-nucleotide tokens for fine-grained resolution, boosting both efficiency and interpretability. Pre-trained on the human genome, NucEL achieves state-of-the-art results on diverse downstream tasks -- regulatory element identification (e.g., promoters, enhancers), transcription factor binding prediction, open chromatin classification, and histone modification profiling -- surpassing similarly sized MLM-based models and rivaling models 25x larger, such as NT. Ablation studies highlight optimal tokenization and masking strategies for ELECTRA-style DNA pre-training. Attention analysis reveals NucEL's superior capture of biologically relevant motifs compared to NT, providing insights into hierarchical learning and regulatory element modeling. These findings demonstrate ELECTRA-style pre-training as an efficient, effective strategy for genomic representation learning with broad implications for genomic research.
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 12:34:51 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Ding",
"Ke",
""
],
[
"Parker",
"Brian",
""
],
[
"Wen",
"Jiayu",
""
]
] |
2508.13192
|
Mingzhe Hu
|
Mingzhe Hu, Zach Eidex, Shansong Wang, Mojtaba Safari, Qiang Li, and Xiaofeng Yang
|
Benchmarking GPT-5 for Zero-Shot Multimodal Medical Reasoning in Radiology and Radiation Oncology
| null | null | null | null |
eess.IV cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Radiology, radiation oncology, and medical physics require decision-making that integrates medical images, textual reports, and quantitative data under high-stakes conditions. With the introduction of GPT-5, it is critical to assess whether recent advances in large multimodal models translate into measurable gains in these safety-critical domains. We present a targeted zero-shot evaluation of GPT-5 and its smaller variants (GPT-5-mini, GPT-5-nano) against GPT-4o across three representative tasks. We present a targeted zero-shot evaluation of GPT-5 and its smaller variants (GPT-5-mini, GPT-5-nano) against GPT-4o across three representative tasks: (1) VQA-RAD, a benchmark for visual question answering in radiology; (2) SLAKE, a semantically annotated, multilingual VQA dataset testing cross-modal grounding; and (3) a curated Medical Physics Board Examination-style dataset of 150 multiple-choice questions spanning treatment planning, dosimetry, imaging, and quality assurance. Across all datasets, GPT-5 achieved the highest accuracy, with substantial gains over GPT-4o up to +20.00% in challenging anatomical regions such as the chest-mediastinal, +13.60% in lung-focused questions, and +11.44% in brain-tissue interpretation. On the board-style physics questions, GPT-5 attained 90.7% accuracy (136/150), exceeding the estimated human passing threshold, while GPT-4o trailed at 78.0%. These results demonstrate that GPT-5 delivers consistent and often pronounced performance improvements over GPT-4o in both image-grounded reasoning and domain-specific numerical problem-solving, highlighting its potential to augment expert workflows in medical imaging and therapeutic physics.
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 16:14:51 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Hu",
"Mingzhe",
""
],
[
"Eidex",
"Zach",
""
],
[
"Wang",
"Shansong",
""
],
[
"Safari",
"Mojtaba",
""
],
[
"Li",
"Qiang",
""
],
[
"Yang",
"Xiaofeng",
""
]
] |
2508.13193
|
Ignacio Chi-Dur\'an
|
Ignacio Chi-Dur\'an, Evan J. Villafranca, David Dang, Rachelle Rosiles, Chun Tung Cheung, Zhiran Zhang, Jason P. Cleveland and Peter C. Maurer
|
Quantum biosensing on a multiplexed functionalized diamond microarray
| null | null | null | null |
physics.ins-det physics.bio-ph physics.chem-ph quant-ph
|
http://creativecommons.org/licenses/by/4.0/
|
Quantum sensing with nitrogen-vacancy (NV) centers in diamond promises to revolutionize biological research and medical diagnostics. Thanks to their high sensitivity, NV sensors could, in principle, detect specific binding events with metabolites and proteins in a massively parallel and label-free way, avoiding the complexity of mass spectrometry. Realizing this vision has been hindered by the lack of quantum sensor arrays that unite high-density spatial multiplexing with uncompromising biochemical specificity. Here, we introduce a scalable quantum biosensing platform that overcomes these barriers by integrating the first multiplexed DNA microarray directly onto a subnanometer antifouling diamond surface. The 7x7 DNA array, patterned onto a diamond chip, enables simultaneous detection of 49 distinct biomolecular features with high spatial resolution and reproducibility, as verified by fluorescence microscopy. Molecular recognition is converted into a quantum signal via a target-induced displacement mechanism in which hybridization removes a Gd$^{3+}$-tagged DNA strand, restoring NV center spin relaxation times (T$_1$) and producing a binary quantum readout. This platform establishes a new paradigm for high-throughput, multiplexed quantum biosensing and opens the door to advanced molecular diagnostics and large-scale quantum sensor networks operable in complex biological environments.
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 18:00:00 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Chi-Durán",
"Ignacio",
""
],
[
"Villafranca",
"Evan J.",
""
],
[
"Dang",
"David",
""
],
[
"Rosiles",
"Rachelle",
""
],
[
"Cheung",
"Chun Tung",
""
],
[
"Zhang",
"Zhiran",
""
],
[
"Cleveland",
"Jason P.",
""
],
[
"Maurer",
"Peter C.",
""
]
] |
2508.13194
|
Benedetto Daniele Militello Dr
|
Benedetto Militello and Anna Napoli
|
Short-time behavior of a system ruled by non-Hermitian time-dependent Hamiltonians
|
10 pages, 5 figures
|
Symmetry 2005, 17(8), 1336
|
10.3390/sym17081336
| null |
quant-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The short-time behavior of the survival probability of a system governed by a time-dependent non-Hermitian Hamiltonian is derived using to the second order perturbative approach. The resulting expression allows for the analysis of some situations which could be of interest in the field of quantum technology. For example, it becomes possible to predict a quantum Zeno effect even in the presence of decay processes.
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 21:05:32 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Militello",
"Benedetto",
""
],
[
"Napoli",
"Anna",
""
]
] |
2508.13195
|
Christopher Pollett
|
Chris Pollett
|
Arithmetics within the Linear Time Hierarchy
| null | null | null | null |
cs.LO math.LO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We identify fragments of the arithmetic $S_1$ that enjoy nice closure properties and have exact characterization of their definable multifunctions. To do this, in the language of $S_1$, $L_1$, starting from the formula classes, $\Sigma^{\mathsf b}_{i}$, which ignore sharply bounded quantifiers when determining quantifier alternations, we define new syntactic classes by counting bounded existential sharply bounded universal quantifiers blocks. Using these, we define arithmetics: $\breve{S}^{i}_{1}$, $TLS^i_1$ and $TSC^i_1$. $\breve{S}^{i}_{1}$ consists of open axioms for the language symbols and length induction for one of our new classes, $SIUT_{i,1}^{\{p(|id|)\}}$. $TLS^i_1$ and $TSC^i_1$ are defined using axioms related to dependent choice sequences for formulas from two other classes within $\Sigma^{\mathsf b}_{i}$. We prove for $i \geq 1$ that $$TLS^i_1 \subseteq TSC^i_1 \subseteq \breve{S}^{i}_{1} \preceq_{\forall B(SITT_{i+1}^{\{p(|id|)\}})} TLS^{i+1}_1$$ and that the $SITT_{i}^{\{p(|id|)\}}$-definable in $TLS^i_1$ (resp. $SITT_{i}^{\{2^{p(||id||)}\}}$-definable in $TSC^i_1$) multifunctions are $L_1$-$FLOGSPACE^{SIT_{i,1}}[wit]$ (resp. $L_1$-$FSC^{SIT_{i,1}}[wit]$). These multifunction classes are respectively the logspace or $SC$ (poly-time, polylog-space) computable multifunctions whose output is bound by a term in $L_1$ and that have access to a witness oracle for another restriction on the $\Sigma^{\mathsf b}_{i}$ formulas, $SIT_{i,1}$. For the $i=1$ cases, this simplifies respectively to the functions in logspace and $SC$, Steve's Class, poly-time, polylog-space. We prove independence results related to the Matiyasevich Robinson Davis Putnam Theorem (MRDP) and to whether our theories prove simultaneous nondeterministic polynomial time, sublinear space is equal to co-nondeterministic polynomial time, sublinear space.
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 21:31:08 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Pollett",
"Chris",
""
]
] |
2508.13196
|
Miloud Mihoubi
|
Meriem Zerkouk, Miloud Mihoubi, Belkacem Chikhaoui
|
Contextual Attention-Based Multimodal Fusion of LLM and CNN for Sentiment Analysis
|
The 38th Canadian Conference on Artificial Intelligence ( 2025 )
| null | null | null |
cs.LG cs.AI cs.IR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This paper introduces a novel approach for multimodal sentiment analysis on social media, particularly in the context of natural disasters, where understanding public sentiment is crucial for effective crisis management. Unlike conventional methods that process text and image modalities separately, our approach seamlessly integrates Convolutional Neural Network (CNN) based image analysis with Large Language Model (LLM) based text processing, leveraging Generative Pre-trained Transformer (GPT) and prompt engineering to extract sentiment relevant features from the CrisisMMD dataset. To effectively model intermodal relationships, we introduce a contextual attention mechanism within the fusion process. Leveraging contextual-attention layers, this mechanism effectively captures intermodality interactions, enhancing the model's comprehension of complex relationships between textual and visual data. The deep neural network architecture of our model learns from these fused features, leading to improved accuracy compared to existing baselines. Experimental results demonstrate significant advancements in classifying social media data into informative and noninformative categories across various natural disasters. Our model achieves a notable 2.43% increase in accuracy and 5.18% in F1-score, highlighting its efficacy in processing complex multimodal data. Beyond quantitative metrics, our approach provides deeper insight into the sentiments expressed during crises. The practical implications extend to real time disaster management, where enhanced sentiment analysis can optimize the accuracy of emergency interventions. By bridging the gap between multimodal analysis, LLM powered text understanding, and disaster response, our work presents a promising direction for Artificial Intelligence (AI) driven crisis management solutions. Keywords:
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 21:34:13 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Zerkouk",
"Meriem",
""
],
[
"Mihoubi",
"Miloud",
""
],
[
"Chikhaoui",
"Belkacem",
""
]
] |
2508.13197
|
Zhiling Zheng
|
Chenru Duan, Aditya Nandy, Shyam Chand Pal, Xin Yang, Wenhao Gao, Yuanqi Du, Hendrik Kra{\ss}, Yeonghun Kang, Varinia Bernales, Zuyang Ye, Tristan Pyle, Ray Yang, Zeqi Gu, Philippe Schwaller, Shengqian Ma, Shijing Sun, Al\'an Aspuru-Guzik, Seyed Mohamad Moosavi, Robert Wexler, Zhiling Zheng
|
The Rise of Generative AI for Metal-Organic Framework Design and Synthesis
|
10 pages, 5 figures
| null | null | null |
cond-mat.mtrl-sci cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Advances in generative artificial intelligence are transforming how metal-organic frameworks (MOFs) are designed and discovered. This Perspective introduces the shift from laborious enumeration of MOF candidates to generative approaches that can autonomously propose and synthesize in the laboratory new porous reticular structures on demand. We outline the progress of employing deep learning models, such as variational autoencoders, diffusion models, and large language model-based agents, that are fueled by the growing amount of available data from the MOF community and suggest novel crystalline materials designs. These generative tools can be combined with high-throughput computational screening and even automated experiments to form accelerated, closed-loop discovery pipelines. The result is a new paradigm for reticular chemistry in which AI algorithms more efficiently direct the search for high-performance MOF materials for clean air and energy applications. Finally, we highlight remaining challenges such as synthetic feasibility, dataset diversity, and the need for further integration of domain knowledge.
|
[
{
"version": "v1",
"created": "Fri, 15 Aug 2025 21:49:17 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Duan",
"Chenru",
""
],
[
"Nandy",
"Aditya",
""
],
[
"Pal",
"Shyam Chand",
""
],
[
"Yang",
"Xin",
""
],
[
"Gao",
"Wenhao",
""
],
[
"Du",
"Yuanqi",
""
],
[
"Kraß",
"Hendrik",
""
],
[
"Kang",
"Yeonghun",
""
],
[
"Bernales",
"Varinia",
""
],
[
"Ye",
"Zuyang",
""
],
[
"Pyle",
"Tristan",
""
],
[
"Yang",
"Ray",
""
],
[
"Gu",
"Zeqi",
""
],
[
"Schwaller",
"Philippe",
""
],
[
"Ma",
"Shengqian",
""
],
[
"Sun",
"Shijing",
""
],
[
"Aspuru-Guzik",
"Alán",
""
],
[
"Moosavi",
"Seyed Mohamad",
""
],
[
"Wexler",
"Robert",
""
],
[
"Zheng",
"Zhiling",
""
]
] |
2508.13198
|
Abicumaran Uthamacumaran
|
Abicumaran Uthamacumaran and Hector Zenil
|
Decoding Neurodevelopmental Dynamics in Pediatric High Grade Gliomas: Divergences in Transcriptomic Network Complexity from Healthy Neural Stem Cell Signatures
| null | null | null | null |
q-bio.QM
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Pediatric high grade gliomas are lethal evolutionary disorders with stalled developmental trajectories and disrupted differentiation hierarchies. We integrate transcriptional and algorithmic network complexity based perturbation analysis to elucidate gene expression patterns and molecular divergence between Diffuse Midline Gliomas and glioblastoma, revealing shared developmental programs steering cell fate decision making. Our complex systems approach supports the emerging paradigm that pediatric high grade gliomas are neurodevelopmental disorders with hybrid lineage identities and disrupted patterning. We identify dysregulated neurodevelopmental and morphogenetic signatures, alongside bioelectric and neurotransmitter signaling programs that alter synaptic organization, neuronal fate commitment, and phenotypic plasticity, regulating glioma phenotypic switching. Causal drivers and regulators of plasticity were predicted as both biomarkers and therapeutic targets, reinforcing the view that pediatric gliomas are disorders of cell fate decisions and collective cellular identity. Decoded plasticity signatures indicate a teleonomic bias toward neural progenitor or neuron like identities, while synaptic transmission gene enrichment supports neuron glioma interactions shaping fate trajectories. These findings advance differentiation therapy as a systems medicine strategy, discovering plasticity regulators to reprogram malignant fates toward stable lineages, offering complex systems based targets for cancer reversion and predictive, preventive, precision oncology.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 00:28:18 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Uthamacumaran",
"Abicumaran",
""
],
[
"Zenil",
"Hector",
""
]
] |
2508.13199
|
Abicumaran Uthamacumaran
|
Abicumaran Uthamacumaran and Hector Zenil
|
Neurosymbolic Learning for Predicting Cell Fate Decisions from Longitudinal Single Cell Transcriptomics in Paediatric Acute Myeloid Leukemia
| null | null | null | null |
q-bio.QM
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Paediatric Acute Myeloid Leukemia is a complex adaptive ecosystem with high morbidity. Current trajectory inference algorithms struggle to predict causal dynamics in AML progression, including relapse and recurrence risk. We propose a symbolic AI and deep learning framework grounded in complexity science, integrating Recurrent Neural Networks, Transformers, and Algorithmic Information Dynamics to model longitudinal single cell transcriptomics and infer complex state transitions in paediatric AML. We identify key plasticity markers as predictive signatures regulating developmental trajectories. These were derived by integrating deep learning with complex systems based network perturbation analysis and dynamical systems theory to infer high dimensional state space attractors steering AML evolution.
Findings reveal dysregulated epigenetic and developmental patterning, with AML cells in maladaptive, reprogrammable plastic states, i.e., developmental arrest blocking terminal differentiation. Predictions forecast neurodevelopmental and morphogenetic signatures guiding AML cell fate bifurcations, suggesting ectoderm mesoderm crosstalk during disrupted differentiation. Neuroplasticity and neurotransmission related transcriptional signals implicate a brain immune hematopoietic axis in AML cell fate cybernetics. This is the first study combining RNNs and AID to predict and decipher longitudinal patterns of cell fate transition trajectories in AML. Our complex systems approach enables causal discovery of predictive biomarkers and therapeutic targets for ecosystem engineering, cancer reversion, precision gene editing, and differentiation therapy, with strong translational potential for precision oncology and patient centered care.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 00:40:56 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Uthamacumaran",
"Abicumaran",
""
],
[
"Zenil",
"Hector",
""
]
] |
2508.13200
|
M Alasli
|
M. Alasli
|
An Intrinsic Barrier for Resolving P = NP (2-SAT as Flat, 3-SAT as High-Dimensional Void-Rich)
| null | null | null | null |
cs.CC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We present a topological barrier to efficient computation, revealed by comparing the geometry of 2 SAT and 3 SAT solution spaces. Viewing the set of satisfying assignments as a cubical complex within the Boolean hypercube, we prove that every 2 SAT instance has a contractible solution space, topologically flat, with all higher Betti numbers bk equals 0 for k greater than or equal 1, while both random and explicit 3 SAT families can exhibit exponential second Betti numbers, corresponding to exponentially many independent voids. These voids are preserved under standard SAT reductions and cannot be collapsed without solving NP-hard subproblems, making them resistant to the three major complexity theoretic barriers, relativization, natural proofs, and algebrization. We establish exponential time lower bounds in restricted query models and extend these to broader algorithmic paradigms under mild information-theoretic or encoding assumptions. This topological contrast flat, connected landscapes in 2 SAT versus tangled, high-dimensional void-rich landscapes in 3 SAT, provides structural evidence toward P does not equal NP, identifying b2 as a paradigm-independent invariant of computational hardness.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 00:49:28 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Alasli",
"M.",
""
]
] |
2508.13201
|
Yang Liu
|
Yang Liu, Lu Zhou, Ruikun He, Rongbo Shen, Yixue Li
|
Benchmarking LLM-based Agents for Single-cell Omics Analysis
| null | null | null | null |
q-bio.GN cs.AI cs.MA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The surge in multimodal single-cell omics data exposes limitations in traditional, manually defined analysis workflows. AI agents offer a paradigm shift, enabling adaptive planning, executable code generation, traceable decisions, and real-time knowledge fusion. However, the lack of a comprehensive benchmark critically hinders progress. We introduce a novel benchmarking evaluation system to rigorously assess agent capabilities in single-cell omics analysis. This system comprises: a unified platform compatible with diverse agent frameworks and LLMs; multidimensional metrics assessing cognitive program synthesis, collaboration, execution efficiency, bioinformatics knowledge integration, and task completion quality; and 50 diverse real-world single-cell omics analysis tasks spanning multi-omics, species, and sequencing technologies. Our evaluation reveals that Grok-3-beta achieves state-of-the-art performance among tested agent frameworks. Multi-agent frameworks significantly enhance collaboration and execution efficiency over single-agent approaches through specialized role division. Attribution analyses of agent capabilities identify that high-quality code generation is crucial for task success, and self-reflection has the most significant overall impact, followed by retrieval-augmented generation (RAG) and planning. This work highlights persistent challenges in code generation, long-context handling, and context-aware knowledge retrieval, providing a critical empirical foundation and best practices for developing robust AI agents in computational biology.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 04:26:18 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Liu",
"Yang",
""
],
[
"Zhou",
"Lu",
""
],
[
"He",
"Ruikun",
""
],
[
"Shen",
"Rongbo",
""
],
[
"Li",
"Yixue",
""
]
] |
2508.13202
|
Kemi Adewale
|
Kemi Y. Adewale, Itegbeyogene P. Ezekiel
|
Magnetic and Mossbauer studies of Ca$_x$Zn$_{1-x}$Fe$_2$O$_4$ nanoferrites
|
7 pages
|
International Journal of Solid State Materials 2019
| null | null |
cond-mat.mtrl-sci
|
http://creativecommons.org/licenses/by/4.0/
|
A one-step synthesis of Ca$_x$Zn$_{1-x}$Fe$_2$O$_4$ (x=0, 0.5 and 1) nanoferrites and nanocomposites by the glycol-thermal method is reported. The structural, morphological and magnetic properties were studied using XRD, HRTEM, HRSEM, Mossbauer spectroscopy and VSM. The XRD patterns show a single phase cubic spinel structure for x=0. A composite phase of a spinel and hematite-like structure was observed for x=0.5, and for x=1, CaFe$_2$O$_4$ has the same structure as hematite. The superparamagnetic nature of the samples was confirmed from the Mossbauer and magnetization results.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 05:39:38 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Adewale",
"Kemi Y.",
""
],
[
"Ezekiel",
"Itegbeyogene P.",
""
]
] |
2508.13203
|
Kemi Adewale
|
Kemi Y. Adewale, Itegbeyogene P.Ezekiel
|
Synthesis and Characterization of Mg doped ZnFe$_2$O$_4$
| null |
Research & Reviews: Journal of Physics 2020
| null | null |
cond-mat.mtrl-sci
|
http://creativecommons.org/licenses/by/4.0/
|
Single-phase Mg-doped ZnFe$_2$O$_4$ nanoparticles with x= 0, 0.3, 0.5, 0.7 have been prepared by the glycol-thermal method without any subsequent calcination. The crystallite size, microstructure and magnetic properties of the prepared nanoparticles were studied using X-ray diffraction (XRD), high resolution scanning electron microscope (TEM), Mossbauer spectroscopy and vibrating sample magnetometer at room temperature. The XRD results revealed the production of a sharp single cubic spinel structure in all the synthesized samples without any impurity peak with the average crystallite size of about 19-28 nm. It was noticed that the lattice parameter varies as the Mg$^{2+}$ ion concentration increases. 57 Mossbauer measurement showed that the nano ferrites exhibit ferrimagnetic and superparamagnetic states. Magnetization measurements confirmed the superparamagnetic behaviour of the samples. The highest coercivity and saturation magnetization were observed at x=0.3. The saturation magnetization (MS) decreases while coercivity (HC) varies with an increase in the concentration of Mg$^{2+}$ ion.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 05:43:43 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Adewale",
"Kemi Y.",
""
],
[
"Ezekiel",
"Itegbeyogene P.",
""
]
] |
2508.13204
|
Dong Liu
|
Dong Liu, Yanxuan Yu
|
QuickMerge++: Fast Token Merging with Autoregressive Prior
|
The paper has been accepted to ICML Tokshop at https://openreview.net/forum?id=dMdxHd0tRf
| null | null | null |
cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
As generative models scale to larger inputs across language, vision, and video domains, the cost of token-level computation has become a key bottleneck. While prior work suggests that only a subset of tokens significantly influence downstream predictions, most token selection methods are static, modality-specific, or incompatible with autoregressive generation. In this paper, we propose QuickMerge, a lightweight token merging framework designed for efficient next-token prediction.
QuickMerge dynamically selects a reduced number of tokens based on attention norm magnitude, guided by an entropy-based budget estimator. To preserve autoregressive compatibility, we introduce a lightweight transformer prior trained over the merged token sequence. By combining semantic salience estimation, flexible token budgets, and AR alignment, QuickMerge enables accurate generation with fewer tokens.
We evaluate QuickMerge across multi-modality domains, demonstrating consistent improvements in compute-accuracy tradeoffs. Specifically, QuickMerge reduces token counts sustantially while matching as well as exceeding the performance of learned tokenizers and fixed-patch baselines.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 06:07:33 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Liu",
"Dong",
""
],
[
"Yu",
"Yanxuan",
""
]
] |
2508.13205
|
Ligang Dong
|
Zhebin Jin and Ligang Dong
|
YOLO11-CR: a Lightweight Convolution-and-Attention Framework for Accurate Fatigue Driving Detection
| null | null | null | null |
cs.CV eess.IV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Driver fatigue detection is of paramount importance for intelligent transportation systems due to its critical role in mitigating road traffic accidents. While physiological and vehicle dynamics-based methods offer accuracy, they are often intrusive, hardware-dependent, and lack robustness in real-world environments. Vision-based techniques provide a non-intrusive and scalable alternative, but still face challenges such as poor detection of small or occluded objects and limited multi-scale feature modeling. To address these issues, this paper proposes YOLO11-CR, a lightweight and efficient object detection model tailored for real-time fatigue detection. YOLO11-CR introduces two key modules: the Convolution-and-Attention Fusion Module (CAFM), which integrates local CNN features with global Transformer-based context to enhance feature expressiveness; and the Rectangular Calibration Module (RCM), which captures horizontal and vertical contextual information to improve spatial localization, particularly for profile faces and small objects like mobile phones. Experiments on the DSM dataset demonstrated that YOLO11-CR achieves a precision of 87.17%, recall of 83.86%, mAP@50 of 88.09%, and mAP@50-95 of 55.93%, outperforming baseline models significantly. Ablation studies further validate the effectiveness of the CAFM and RCM modules in improving both sensitivity and localization accuracy. These results demonstrate that YOLO11-CR offers a practical and high-performing solution for in-vehicle fatigue monitoring, with strong potential for real-world deployment and future enhancements involving temporal modeling, multi-modal data integration, and embedded optimization.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 07:19:04 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Jin",
"Zhebin",
""
],
[
"Dong",
"Ligang",
""
]
] |
2508.13206
|
Ghazal Jamirad
|
Ghazal Jamirad, Mahdi Seif, Abbas Montazeri
|
Fine-Tuning Hydrophilic-Hydrophobic Balance in Stimuli-Responsive PEGPNIPAM Micelles for Controlled Drug Delivery
| null |
Journal of Molecular Liquids, 436, 2025
|
10.1016/j.molliq.2025.128239
| null |
physics.med-ph cond-mat.mtrl-sci
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Temperature-responsive polymeric micelles have great potential in drug delivery, enhancing therapeutic efficacy while minimizing systemic toxicity. This study investigates the self-assembly of poly(ethylene glycol)-poly(N-isopropylacrylamide) (PEG-PNIPAM) block copolymers as nanocarriers for Doxorubicin (DOX) using coarse-grained molecular dynamics (CG-MD) simulations. We systematically examined the effects of polymer chain length and hydrophilic/hydrophobic balance on micelle formation, stability, and drug encapsulation efficiency. PEG-PNIPAM copolymers self-assembled above the lower critical solution temperature (LCST), forming stable micelles with varying morphologies. Our findings revealed that PEG enhances micelle hydration and stability, while PNIPAM improves drug retention but may hinder controlled release. Increasing PNIPAM or PEG length significantly increased micelle size, with DOX preferentially localizing in the PEG-rich outer shell, enhancing solubility and preventing aggregation. The DOX diffusion coefficient ranged from $10^{-8}$ to $10^{-9}$ cm$^{2}$s$^{-1}$, reflecting drug release dynamics. Structural and thermodynamic analyses confirmed spontaneous DOX encapsulation, with PEG-PNIPAM micelles providing a favorable environment for drug loading. Mean square displacement (MSD), radial distribution function (RDF), and solvent-accessible surface area (SASA) analyses highlighted the role of PEG in controlled drug release and PNIPAM in drug binding. These molecular-level insights underscore the potential of fine-tuning PEG-PNIPAM compositions for enhanced drug delivery, paving the way for precision cancer therapies.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 07:56:18 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Jamirad",
"Ghazal",
""
],
[
"Seif",
"Mahdi",
""
],
[
"Montazeri",
"Abbas",
""
]
] |
2508.13207
|
Ali Akbar Kiaei Khoshroudbari
|
S. Z. Pirasteh, Ali A. Kiaei, Mahnaz Bush, Sabra Moghadam, Raha Aghaei, Behnaz Sadeghigol
|
Utilizing the RAIN method and Graph SAGE Model to Identify Effective Drug Combinations for Gastric Neoplasm Treatment
|
43 pages
| null | null | null |
q-bio.QM cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Background: Gastric neoplasm, primarily adenocarcinoma, is an aggressive cancer with high mortality, often diagnosed late, leading to complications like metastasis. Effective drug combinations are vital to address disease heterogeneity, enhance efficacy, reduce resistance, and improve patient outcomes. Methods: The RAIN method integrated Graph SAGE to propose drug combinations, using a graph model with p-value-weighted edges connecting drugs, genes, and proteins. NLP and systematic literature review (PubMed, Scopus, etc.) validated proposed drugs, followed by network meta-analysis to assess efficacy, implemented in Python. Results: Oxaliplatin, fluorouracil, and trastuzumab were identified as effective, supported by 61 studies. Fluorouracil alone had a p-value of 0.0229, improving to 0.0099 with trastuzumab, and 0.0069 for the triple combination, indicating superior efficacy. Conclusion: The RAIN method, combining AI and network meta-analysis, effectively identifies optimal drug combinations for gastric neoplasm, offering a promising strategy to enhance treatment outcomes and guide health policy.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 10:19:19 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Pirasteh",
"S. Z.",
""
],
[
"Kiaei",
"Ali A.",
""
],
[
"Bush",
"Mahnaz",
""
],
[
"Moghadam",
"Sabra",
""
],
[
"Aghaei",
"Raha",
""
],
[
"Sadeghigol",
"Behnaz",
""
]
] |
2508.13208
|
Guohong Peng
|
Yiyang Jia, Zheng Wei, Zheng Yang, Guohong Peng
|
Modeling GRNs with a Probabilistic Categorical Framework
|
21 pages, 5 figures
| null | null | null |
q-bio.MN cs.LG math.CT
|
http://creativecommons.org/licenses/by/4.0/
|
Understanding the complex and stochastic nature of Gene Regulatory Networks (GRNs) remains a central challenge in systems biology. Existing modeling paradigms often struggle to effectively capture the intricate, multi-factor regulatory logic and to rigorously manage the dual uncertainties of network structure and kinetic parameters. In response, this work introduces the Probabilistic Categorical GRN(PC-GRN) framework. It is a novel theoretical approach founded on the synergistic integration of three core methodologies. Firstly, category theory provides a formal language for the modularity and composition of regulatory pathways. Secondly, Bayesian Typed Petri Nets (BTPNs) serve as an interpretable,mechanistic substrate for modeling stochastic cellular processes, with kinetic parameters themselves represented as probability distributions. The central innovation of PC-GRN is its end-to-end generative Bayesian inference engine, which learns a full posterior distribution over BTPN models (P (G, {\Theta}|D)) directly from data. This is achieved by the novel interplay of a GFlowNet, which learns a policy to sample network topologies, and a HyperNetwork, which performs amortized inference to predict their corresponding parameter distributions. The resulting framework provides a mathematically rigorous, biologically interpretable, and uncertainty-aware representation of GRNs, advancing predictive modeling and systems-level analysis.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 14:06:53 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Jia",
"Yiyang",
""
],
[
"Wei",
"Zheng",
""
],
[
"Yang",
"Zheng",
""
],
[
"Peng",
"Guohong",
""
]
] |
2508.13209
|
Yaying Luo
|
Yaying Luo, Hui Fang, Zhu Sun
|
Research on Conversational Recommender System Considering Consumer Types
|
10 pages
| null | null | null |
cs.IR cs.AI cs.SI
|
http://creativecommons.org/licenses/by/4.0/
|
Conversational Recommender Systems (CRS) provide personalized services through multi-turn interactions, yet most existing methods overlook users' heterogeneous decision-making styles and knowledge levels, which constrains both accuracy and efficiency. To address this gap, we propose CT-CRS (Consumer Type-Enhanced Conversational Recommender System), a framework that integrates consumer type modeling into dialogue recommendation. Based on consumer type theory, we define four user categories--dependent, efficient, cautious, and expert--derived from two dimensions: decision-making style (maximizers vs. satisficers) and knowledge level (high vs. low). CT-CRS employs interaction histories and fine-tunes the large language model to automatically infer user types in real time, avoiding reliance on static questionnaires. We incorporate user types into state representation and design a type-adaptive policy that dynamically adjusts recommendation granularity, diversity, and attribute query complexity. To further optimize the dialogue policy, we adopt Inverse Reinforcement Learning (IRL), enabling the agent to approximate expert-like strategies conditioned on consumer type. Experiments on LastFM, Amazon-Book, and Yelp show that CTCRS improves recommendation success rate and reduces interaction turns compared to strong baselines. Ablation studies confirm that both consumer type modeling and IRL contribute significantly to performance gains. These results demonstrate that CT-CRS offers a scalable and interpretable solution for enhancing CRS personalization through the integration of psychological modeling and advanced policy optimization.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 15:15:52 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Luo",
"Yaying",
""
],
[
"Fang",
"Hui",
""
],
[
"Sun",
"Zhu",
""
]
] |
2508.13210
|
Kumar Abhishek
|
Kumar Abhishek
|
Strongly Set-Colorable Graphs: A Complete Characterization
|
5 pages
| null | null | null |
math.CO
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
In this note, we revisit the notion of strong set-colorings introduced by Hegde (2009) and completed by equivalences due to Boutin et al. (2010) and provide a necessary and sufficient \emph{Steiner packing} characterisation: a finite graph $G$ is strongly set-colorable if and only if its associated $3$-uniform hypergraph is a $(2,3,2^{n}-1)$-packing of the unique Steiner triple system $S(2,3,2^{n}-1)$. This unification allows many earlier necessary conditions to be derived instantly as corollaries, streamlining the structure theory of strongly set-colorable graphs.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 19:38:40 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Abhishek",
"Kumar",
""
]
] |
2508.13211
|
Jose Pereira Frugone A
|
Jose Alberto Pereira Frugone
|
Quantum geometric phase with initial state index dependent on space-time curvature
|
7 pages
| null | null | null |
quant-ph gr-qc
|
http://creativecommons.org/licenses/by/4.0/
|
We introduce a new way of generating the quantum geometric phase by making the initial base state index dependent on space-time curvature. We prove that the resulting Schr\"odinger equation is identical to the trace form of the Einstein gravitational field equations when we are in the adiabatic approximation. This particular initial condition was proposed in a previous work by the author.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 19:46:39 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Frugone",
"Jose Alberto Pereira",
""
]
] |
2508.13212
|
Takfarinas Medani
|
Takfarinas Medani, Jace Willis, Chris Wright, Yash Vakilna, Ryan Shores, Raymundo Cassani, Anand Joshi, Richard Leahy, John Seymour, and John Mosher
|
High-Resolution Directional Depth Electrodes: Open-Source FEM Lead-Field Modeling, Characterization, and Validation
|
20 pages, 12 figures
| null | null | null |
physics.med-ph
|
http://creativecommons.org/licenses/by/4.0/
|
Depth electrodes used in stereoelectroencephalography (sEEG) and deep-brain stimulation (DBS) are essential tools for neural recording and stimulation. Traditional designs have limited spatial resolution, typically 8 to 16 cylindrical contacts (0.8 to 1.0 mm diameter) along a 5 to 10 cm shaft, restricting recordings from small or localized populations. Recent high-density, directional electrodes (HDsEEG) enable finer localization of local field potentials (LFPs) and spike timing. Yet, characterizing their directional sensitivity and validating modeling tools for lead field (LF) analysis remain critical. We compare finite element method (FEM) LF modeling of a novel HD-sEEG electrode using two tools: a commercial solver (ANSYS) and an open-source pipeline (Brainstorm-DUNEuro). Goals: (i) validate against analytical solutions, (ii) assess solver differences, and (iii) characterize HDsEEG directional sensitivity. LFs were modeled in simple and bio-relevant scenarios. Using Helmholtz reciprocity, we computed LFs by (a) electrode-based stimulation with ANSYS and (b) source-based recording with Brainstorm-DUNEuro. First, a multi-sphere head model with known solutions tested the solver's accuracy. Next, an HDsEEG electrode in a homogeneous conductor was simulated. Directional effects were assessed by comparing sensitivity with vs. without the insulating substrate. Source localization performance was also compared between HDsEEG and standard electrodes. Both solvers closely matched analytic solutions. In realistic settings, LF distributions were highly similar. Modeling showed clear directional sensitivity: contacts facing a source had higher sensitivity than those shadowed, reflecting a substrate shielding effect that vanished when the substrate was conductive. Crucially, HDsEEG improved source localization, as voltage differences across contacts provided robust directional LF information.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 22:31:42 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Medani",
"Takfarinas",
""
],
[
"Willis",
"Jace",
""
],
[
"Wright",
"Chris",
""
],
[
"Vakilna",
"Yash",
""
],
[
"Shores",
"Ryan",
""
],
[
"Cassani",
"Raymundo",
""
],
[
"Joshi",
"Anand",
""
],
[
"Leahy",
"Richard",
""
],
[
"Seymour",
"John",
""
],
[
"Mosher",
"John",
""
]
] |
2508.13213
|
Adamo Cerioli
|
Adamo Cerioli, Edward D. Lee, Vito D. P. Servedio
|
AI sustains higher strategic tension than humans in chess
| null | null | null | null |
cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Strategic decision-making involves managing the tension between immediate opportunities and long-term objectives. We study this trade-off in chess by characterizing and comparing dynamics between human vs human and AI vs AI games. We propose a network-based metric of piece-to-piece interaction to quantify the ongoing strategic tension on the board. Its evolution in games reveals that the most competitive AI players sustain higher levels of strategic tension for longer durations than elite human players. Cumulative tension varies with algorithmic complexity for AI and correspondingly in human-played games increases abruptly with expertise at about 1600 Elo and again at 2300 Elo. The profiles reveal different approaches. Highly competitive AI tolerates interconnected positions balanced between offensive and defensive tactics over long periods. Human play, in contrast, limits tension and game complexity, which may reflect cognitive limitations and adaptive strategies. The difference may have implications for AI usage in complex, strategic environments.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 22:53:34 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Cerioli",
"Adamo",
""
],
[
"Lee",
"Edward D.",
""
],
[
"Servedio",
"Vito D. P.",
""
]
] |
2508.13214
|
Zekai Huang
|
Xuyang Guo, Zekai Huang, Zhao Song, Jiahao Zhang
|
Too Easily Fooled? Prompt Injection Breaks LLMs on Frustratingly Simple Multiple-Choice Questions
| null | null | null | null |
cs.CR cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Large Language Models (LLMs) have recently demonstrated strong emergent abilities in complex reasoning and zero-shot generalization, showing unprecedented potential for LLM-as-a-judge applications in education, peer review, and data quality evaluation. However, their robustness under prompt injection attacks, where malicious instructions are embedded into the content to manipulate outputs, remains a significant concern. In this work, we explore a frustratingly simple yet effective attack setting to test whether LLMs can be easily misled. Specifically, we evaluate LLMs on basic arithmetic questions (e.g., "What is 3 + 2?") presented as either multiple-choice or true-false judgment problems within PDF files, where hidden prompts are injected into the file. Our results reveal that LLMs are indeed vulnerable to such hidden prompt injection attacks, even in these trivial scenarios, highlighting serious robustness risks for LLM-as-a-judge applications.
|
[
{
"version": "v1",
"created": "Sat, 16 Aug 2025 23:30:08 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Guo",
"Xuyang",
""
],
[
"Huang",
"Zekai",
""
],
[
"Song",
"Zhao",
""
],
[
"Zhang",
"Jiahao",
""
]
] |
2508.13216
|
Toni Schneidereit
|
Santosh Humagain, Toni Schneidereit
|
Strategies for training point distributions in physics-informed neural networks
| null | null | null | null |
cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Physics-informed neural networks approach the approximation of differential equations by directly incorporating their structure and given conditions in a loss function. This enables conditions like, e.g., invariants to be easily added during the modelling phase. In addition, the approach can be considered as mesh free and can be utilised to compute solutions on arbitrary grids after the training phase. Therefore, physics-informed neural networks are emerging as a promising alternative to solving differential equations with methods from numerical mathematics. However, their performance highly depends on a large variety of factors. In this paper, we systematically investigate and evaluate a core component of the approach, namely the training point distribution. We test two ordinary and two partial differential equations with five strategies for training data generation and shallow network architectures, with one and two hidden layers. In addition to common distributions, we introduce sine-based training points, which are motivated by the construction of Chebyshev nodes. The results are challenged by using certain parameter combinations like, e.g., random and fixed-seed weight initialisation for reproducibility. The results show the impact of the training point distributions on the solution accuracy and we find evidence that they are connected to the characteristics of the differential equation.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 09:40:49 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Humagain",
"Santosh",
""
],
[
"Schneidereit",
"Toni",
""
]
] |
2508.13217
|
Di Hu
|
Di Hu (1), Yawen Guo (1), Ha Na Cho (1), Emilie Chow (1), Dana B. Mukamel (1), Dara Sorkin (1), Andrew Reikes (1), Danielle Perret (1), Deepti Pandita (1), Kai Zheng (1) ((1) University of California, Irvine, Irvine, CA, USA)
|
When AI Writes Back: Ethical Considerations by Physicians on AI-Drafted Patient Message Replies
|
Paper accepted for the proceedings of the 2025 American Medical Informatics Association Annual Symposium (AMIA)
| null | null | null |
cs.CY cs.HC
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
The increasing burden of responding to large volumes of patient messages has become a key factor contributing to physician burnout. Generative AI (GenAI) shows great promise to alleviate this burden by automatically drafting patient message replies. The ethical implications of this use have however not been fully explored. To address this knowledge gap, we conducted a semi-structured interview study with 21 physicians who participated in a GenAI pilot program. We found that notable ethical considerations expressed by the physician participants included human oversight as ethical safeguard, transparency and patient consent of AI use, patient misunderstanding of AI's role, and patient privacy and data security as prerequisites. Additionally, our findings suggest that the physicians believe the ethical responsibility of using GenAI in this context primarily lies with users, not with the technology. These findings may provide useful insights into guiding the future implementation of GenAI in clinical practice.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 09:46:49 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Hu",
"Di",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Guo",
"Yawen",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Cho",
"Ha Na",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Chow",
"Emilie",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Mukamel",
"Dana B.",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Sorkin",
"Dara",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Reikes",
"Andrew",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Perret",
"Danielle",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Pandita",
"Deepti",
"",
"University of California, Irvine, Irvine, CA, USA"
],
[
"Zheng",
"Kai",
"",
"University of California, Irvine, Irvine, CA, USA"
]
] |
2508.13218
|
Frederik Baucks
|
Frederik Baucks, Robin Schmucker, Laurenz Wiskott
|
The Course Difficulty Analysis Cookbook
| null | null | null | null |
cs.CY cs.LG stat.AP
|
http://creativecommons.org/licenses/by/4.0/
|
Curriculum analytics (CA) studies curriculum structure and student data to ensure the quality of educational programs. An essential aspect is studying course properties, which involves assigning each course a representative difficulty value. This is critical for several aspects of CA, such as quality control (e.g., monitoring variations over time), course comparisons (e.g., articulation), and course recommendation (e.g., advising). Measuring course difficulty requires careful consideration of multiple factors: First, when difficulty measures are sensitive to the performance level of enrolled students, it can bias interpretations by overlooking student diversity. By assessing difficulty independently of enrolled students' performances, we can reduce the risk of bias and enable fair, representative assessments of difficulty. Second, from a measurement theoretic perspective, the measurement must be reliable and valid to provide a robust basis for subsequent analyses. Third, difficulty measures should account for covariates, such as the characteristics of individual students within a diverse populations (e.g., transfer status). In recent years, various notions of difficulty have been proposed. This paper provides the first comprehensive review and comparison of existing approaches for assessing course difficulty based on grade point averages and latent trait modeling. It further offers a hands-on tutorial on model selection, assumption checking, and practical CA applications. These applications include monitoring course difficulty over time and detecting courses with disparate outcomes between distinct groups of students (e.g., dropouts vs. graduates), ultimately aiming to promote high-quality, fair, and equitable learning experiences. To support further research and application, we provide an open-source software package and artificial datasets, facilitating reproducibility and adoption.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 10:55:37 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Baucks",
"Frederik",
""
],
[
"Schmucker",
"Robin",
""
],
[
"Wiskott",
"Laurenz",
""
]
] |
2508.13219
|
Xiaohua Qi
|
Su Chen, Xiaohua Qi, Xixun Lin, Yanmin Shang, Xiaolin Xu and Yangxi Li
|
Deep Graph Neural Point Process For Learning Temporal Interactive Networks
| null | null | null | null |
cs.LG cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Learning temporal interaction networks(TIN) is previously regarded as a coarse-grained multi-sequence prediction problem, ignoring the network topology structure influence. This paper addresses this limitation and a Deep Graph Neural Point Process(DGNPP) model for TIN is proposed. DGNPP consists of two key modules: the Node Aggregation Layer and the Self Attentive Layer. The Node Aggregation Layer captures topological structures to generate static representation for users and items, while the Self Attentive Layer dynamically updates embeddings over time. By incorporating both dynamic and static embeddings into the event intensity function and optimizing the model via maximum likelihood estimation, DGNPP predicts events and occurrence time effectively. Experimental evaluations on three public datasets demonstrate that DGNPP achieves superior performance in event prediction and time prediction tasks with high efficiency, significantly outperforming baseline models and effectively mitigating the limitations of prior approaches.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 11:17:03 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Chen",
"Su",
""
],
[
"Qi",
"Xiaohua",
""
],
[
"Lin",
"Xixun",
""
],
[
"Shang",
"Yanmin",
""
],
[
"Xu",
"Xiaolin",
""
],
[
"Li",
"Yangxi",
""
]
] |
2508.13220
|
Daoyuan Wu
|
Yixuan Yang and Daoyuan Wu and Yufan Chen
|
MCPSecBench: A Systematic Security Benchmark and Playground for Testing Model Context Protocols
|
This is a technical report from Lingnan University, Hong Kong
| null | null | null |
cs.CR cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Large Language Models (LLMs) are increasingly integrated into real-world applications via the Model Context Protocol (MCP), a universal, open standard for connecting AI agents with data sources and external tools. While MCP enhances the capabilities of LLM-based agents, it also introduces new security risks and expands their attack surfaces. In this paper, we present the first systematic taxonomy of MCP security, identifying 17 attack types across 4 primary attack surfaces. We introduce MCPSecBench, a comprehensive security benchmark and playground that integrates prompt datasets, MCP servers, MCP clients, and attack scripts to evaluate these attacks across three major MCP providers. Our benchmark is modular and extensible, allowing researchers to incorporate custom implementations of clients, servers, and transport protocols for systematic security assessment. Experimental results show that over 85% of the identified attacks successfully compromise at least one platform, with core vulnerabilities universally affecting Claude, OpenAI, and Cursor, while prompt-based and tool-centric attacks exhibit considerable variability across different hosts and models. Overall, MCPSecBench standardizes the evaluation of MCP security and enables rigorous testing across all MCP layers.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 11:49:16 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Yang",
"Yixuan",
""
],
[
"Wu",
"Daoyuan",
""
],
[
"Chen",
"Yufan",
""
]
] |
2508.13221
|
Neda Farhangkhah
|
N. Farhangkhah, S. Hajkhalili
|
Phase Transitions of Electromagnetically Charged Black Holes in Lovelock Gravity with Nonconstant Curvature Horizons
|
18 pages, 6 figures
| null | null | null |
gr-qc hep-th
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We present the most general class of charged black hole solutions in third-order Lovelock gravity within even-dimensional spacetimes in the presence of an electromagnetic field. These solutions feature nonconstant-curvature horizons that affect geometry when n>=8. The near-origin behavior of the metric reveals a timelike singularity for electrically charged cases, in contrast to the spacelike singularity found in the uncharged case. We investigate thermodynamic stability in both the grand canonical and canonical ensembles. In the grand canonical ensemble, stability is determined by the positivity of both the Hessian determinant and the temperature. In the canonical ensemble, the sign of the heat capacity governs stability. We identify both first- and second-order phase transitions, including a van der Waals-like behavior characterized by instability at intermediate black hole sizes. Our results reveal a rich phase structure influenced by Lovelock corrections and electromagnetic fields, and demonstrate how conserved charges affect black hole evaporation and stabilization.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 11:50:09 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Farhangkhah",
"N.",
""
],
[
"Hajkhalili",
"S.",
""
]
] |
2508.13222
|
Faranak Farshadifar
|
F. Farshadifar
|
Some results on the ideal-based cozero-divisor graph of a commutative ring
| null | null | null | null |
math.AC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Let R be a commutative ring with identity and I be an ideal of R. The cozero-divisor graph with respect to I, denoted by $\Gamma''_I(R)$, is the graph of R with vertices {x \in R -I :xR +I \not=R} and two distinct vertices $x$ and $y$ are adjacent if and only if $x \not \in yR+I$ and $y \not \in xR+I$.In this paper, we obtained some results on $\Gamma''_I(R)$.
|
[
{
"version": "v1",
"created": "Sun, 17 Aug 2025 12:47:01 GMT"
}
] | 2025-08-20T00:00:00 |
[
[
"Farshadifar",
"F.",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.