id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
list |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2508.10298
|
Weijian Mai
|
Weijian Mai, Jiamin Wu, Yu Zhu, Zhouheng Yao, Dongzhan Zhou, Andrew F. Luo, Qihao Zheng, Wanli Ouyang, Chunfeng Song
|
SynBrain: Enhancing Visual-to-fMRI Synthesis via Probabilistic Representation Learning
| null | null | null | null |
cs.LG cs.CV eess.IV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Deciphering how visual stimuli are transformed into cortical responses is a fundamental challenge in computational neuroscience. This visual-to-neural mapping is inherently a one-to-many relationship, as identical visual inputs reliably evoke variable hemodynamic responses across trials, contexts, and subjects. However, existing deterministic methods struggle to simultaneously model this biological variability while capturing the underlying functional consistency that encodes stimulus information. To address these limitations, we propose SynBrain, a generative framework that simulates the transformation from visual semantics to neural responses in a probabilistic and biologically interpretable manner. SynBrain introduces two key components: (i) BrainVAE models neural representations as continuous probability distributions via probabilistic learning while maintaining functional consistency through visual semantic constraints; (ii) A Semantic-to-Neural Mapper acts as a semantic transmission pathway, projecting visual semantics into the neural response manifold to facilitate high-fidelity fMRI synthesis. Experimental results demonstrate that SynBrain surpasses state-of-the-art methods in subject-specific visual-to-fMRI encoding performance. Furthermore, SynBrain adapts efficiently to new subjects with few-shot data and synthesizes high-quality fMRI signals that are effective in improving data-limited fMRI-to-image decoding performance. Beyond that, SynBrain reveals functional consistency across trials and subjects, with synthesized signals capturing interpretable patterns shaped by biological neural variability. The code will be made publicly available.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:01:05 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Mai",
"Weijian",
""
],
[
"Wu",
"Jiamin",
""
],
[
"Zhu",
"Yu",
""
],
[
"Yao",
"Zhouheng",
""
],
[
"Zhou",
"Dongzhan",
""
],
[
"Luo",
"Andrew F.",
""
],
[
"Zheng",
"Qihao",
""
],
[
"Ouyang",
"Wanli",
""
],
[
"Song",
"Chunfeng",
""
]
] |
2508.10299
|
Danni Peng
|
Danni Peng, Yuan Wang, Kangning Cai, Peiyan Ning, Jiming Xu, Yong Liu, Rick Siow Mong Goh, Qingsong Wei, Huazhu Fu
|
Improving Learning of New Diseases through Knowledge-Enhanced Initialization for Federated Adapter Tuning
| null | null | null | null |
cs.LG cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In healthcare, federated learning (FL) is a widely adopted framework that enables privacy-preserving collaboration among medical institutions. With large foundation models (FMs) demonstrating impressive capabilities, using FMs in FL through cost-efficient adapter tuning has become a popular approach. Given the rapidly evolving healthcare environment, it is crucial for individual clients to quickly adapt to new tasks or diseases by tuning adapters while drawing upon past experiences. In this work, we introduce Federated Knowledge-Enhanced Initialization (FedKEI), a novel framework that leverages cross-client and cross-task transfer from past knowledge to generate informed initializations for learning new tasks with adapters. FedKEI begins with a global clustering process at the server to generalize knowledge across tasks, followed by the optimization of aggregation weights across clusters (inter-cluster weights) and within each cluster (intra-cluster weights) to personalize knowledge transfer for each new task. To facilitate more effective learning of the inter- and intra-cluster weights, we adopt a bi-level optimization scheme that collaboratively learns the global intra-cluster weights across clients and optimizes the local inter-cluster weights toward each client's task objective. Extensive experiments on three benchmark datasets of different modalities, including dermatology, chest X-rays, and retinal OCT, demonstrate FedKEI's advantage in adapting to new diseases compared to state-of-the-art methods.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:02:48 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Peng",
"Danni",
""
],
[
"Wang",
"Yuan",
""
],
[
"Cai",
"Kangning",
""
],
[
"Ning",
"Peiyan",
""
],
[
"Xu",
"Jiming",
""
],
[
"Liu",
"Yong",
""
],
[
"Goh",
"Rick Siow Mong",
""
],
[
"Wei",
"Qingsong",
""
],
[
"Fu",
"Huazhu",
""
]
] |
2508.10300
|
Kunal Menda
|
Kunal Menda and Raphael S Benarrosh
|
Optimal Capital Deployment Under Stochastic Deal Arrivals: A Continuous-Time ADP Approach
| null | null | null | null |
q-fin.PM
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Suppose you are a fund manager with \$100 million to deploy and two years to invest it. A deal comes across your desk that looks appealing but costs \$50 million -- half of your available capital. Should you take it, or wait for something better? The decision hinges on the trade-off between current opportunities and uncertain future arrivals. This work formulates the problem of capital deployment under stochastic deal arrivals as a continuous-time Markov decision process (CTMDP) and solves it numerically via an approximate dynamic programming (ADP) approach. We model deal economics using correlated lognormal distributions for multiples on invested capital (MOIC) and deal sizes, and model arrivals as a nonhomogeneous Poisson process (NHPP). Our approach uses quasi-Monte Carlo (QMC) sampling to efficiently approximate the continuous-time Bellman equation for the value function over a discretized capital grid. We present an interpretable acceptance policy, illustrating how selectivity evolves over time and as capital is consumed. We show in simulation that this policy outperforms a baseline that accepts any affordable deal exceeding a fixed hurdle rate.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:05:15 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Menda",
"Kunal",
""
],
[
"Benarrosh",
"Raphael S",
""
]
] |
2508.10301
|
Shao-Ming Fei
|
Zong Wang, Zhihao Ma, Lin Chen, Chengjie Zhang, Shao-Ming Fei
|
Unified Construction of Genuine Multipartite Entanglement Measures Based on Geometric Mean and its Applications
|
8 pages, 3 figures
|
Quant. Inform. Processing 24 (2025) 252
| null | null |
quant-ph
|
http://creativecommons.org/licenses/by/4.0/
|
Genuine multipartite entanglement (GME) is an important resource in quantum information processing. We systematically study the measures of GME based on the geometric mean of bi-partition entanglements and present a unified construction of GME measures, which gives rise to the widely used GME measures including GME concurrence, the convex-roof extended negativity of GME, the geometric measure of entanglement of GME. Our GME measures satisfy the desirable conditions such as scalability and smoothness. Moreover, we provide fidelity-based analytical lower bounds for our GME measures. Our bounds are tight and can be estimated experiment friendly without requiring quantum state tomography. Furthermore, we apply our results to study the dynamics of GME. We identify an initial condition that influences the sudden death of genuine quadripartite entanglement under individual non-Markovian processes. The GME of Dirac particles with Hawking radiation in the background of a Schwarzschild black hole is also investigated.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:09:26 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Wang",
"Zong",
""
],
[
"Ma",
"Zhihao",
""
],
[
"Chen",
"Lin",
""
],
[
"Zhang",
"Chengjie",
""
],
[
"Fei",
"Shao-Ming",
""
]
] |
2508.10302
|
Xun Lu
|
Xun Lu and Liangjun Su
|
Two-Way Mean Group Estimators for Heterogeneous Panel Models with Fixed T
| null | null | null | null |
econ.EM
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We consider a correlated random coefficient panel data model with two-way fixed effects and interactive fixed effects in a fixed T framework. We propose a two-way mean group (TW-MG) estimator for the expected value of the slope coefficient and propose a leave-one-out jackknife method for valid inference. We also consider a pooled estimator and provide a Hausman-type test for poolability. Simulations demonstrate the excellent performance of our estimators and inference methods in finite samples. We apply our new methods to two datasets to examine the relationship between health-care expenditure and income, and estimate a production function.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:15:02 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Lu",
"Xun",
""
],
[
"Su",
"Liangjun",
""
]
] |
2508.10303
|
Arkapravo Ghosh
|
Arkapravo Ghosh, Abhishek Moitra, Abhiroop Bhattacharjee, Ruokai Yin and Priyadarshini Panda
|
DiffAxE: Diffusion-driven Hardware Accelerator Generation and Design Space Exploration
|
14 pages, 24 figures, 8 tables, 54 references
| null | null | null |
cs.AR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Design space exploration (DSE) is critical for developing optimized hardware architectures, especially for AI workloads such as deep neural networks (DNNs) and large language models (LLMs), which require specialized acceleration. As model complexity grows, accelerator design spaces have expanded to O(10^17), becoming highly irregular, non-convex, and exhibiting many-to-one mappings from design configurations to performance metrics. This complexity renders direct inverse derivation infeasible and necessitates heuristic or sampling-based optimization. Conventional methods - including Bayesian optimization, gradient descent, reinforcement learning, and genetic algorithms - depend on iterative sampling, resulting in long runtimes and sensitivity to initialization. Deep learning-based approaches have reframed DSE as classification using recommendation models, but remain limited to small-scale (O(10^3)), less complex design spaces. To overcome these constraints, we propose a generative approach that models hardware design as 1-D image synthesis conditioned on target performance, enabling efficient learning of non-differentiable, non-bijective hardware-performance mappings. Our framework achieves 0.86% lower generation error than Bayesian optimization with a 17000x speedup, and outperforms GANDSE with 30% lower error at only 1.83x slower search. We further extend the method to a structured DSE setting, attaining 9.8% lower energy-delay product (EDP) and 6% higher performance, with up to 145.6x and 1312x faster search compared to existing optimization methods on O(10^17) design spaces. For LLM inference, our method achieves 3.37x and 7.75x lower EDP on a 32nm ASIC and Xilinx Ultrascale+ VPU13 FPGA, respectively, compared to the state-of-the-art DOSA framework.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:19:34 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Ghosh",
"Arkapravo",
""
],
[
"Moitra",
"Abhishek",
""
],
[
"Bhattacharjee",
"Abhiroop",
""
],
[
"Yin",
"Ruokai",
""
],
[
"Panda",
"Priyadarshini",
""
]
] |
2508.10304
|
Sandra Avila
|
Gustavo Bonil, Simone Hashiguti, Jhessica Silva, Jo\~ao Gondim, Helena Maia, N\'adia Silva, Helio Pedrini, Sandra Avila
|
Yet another algorithmic bias: A Discursive Analysis of Large Language Models Reinforcing Dominant Discourses on Gender and Race
|
29 pages, 3 figures
| null | null | null |
cs.CL cs.AI
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
With the advance of Artificial Intelligence (AI), Large Language Models (LLMs) have gained prominence and been applied in diverse contexts. As they evolve into more sophisticated versions, it is essential to assess whether they reproduce biases, such as discrimination and racialization, while maintaining hegemonic discourses. Current bias detection approaches rely mostly on quantitative, automated methods, which often overlook the nuanced ways in which biases emerge in natural language. This study proposes a qualitative, discursive framework to complement such methods. Through manual analysis of LLM-generated short stories featuring Black and white women, we investigate gender and racial biases. We contend that qualitative methods such as the one proposed here are fundamental to help both developers and users identify the precise ways in which biases manifest in LLM outputs, thus enabling better conditions to mitigate them. Results show that Black women are portrayed as tied to ancestry and resistance, while white women appear in self-discovery processes. These patterns reflect how language models replicate crystalized discursive representations, reinforcing essentialization and a sense of social immobility. When prompted to correct biases, models offered superficial revisions that maintained problematic meanings, revealing limitations in fostering inclusive narratives. Our results demonstrate the ideological functioning of algorithms and have significant implications for the ethical use and development of AI. The study reinforces the need for critical, interdisciplinary approaches to AI design and deployment, addressing how LLM-generated discourses reflect and perpetuate inequalities.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:22:02 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Bonil",
"Gustavo",
""
],
[
"Hashiguti",
"Simone",
""
],
[
"Silva",
"Jhessica",
""
],
[
"Gondim",
"João",
""
],
[
"Maia",
"Helena",
""
],
[
"Silva",
"Nádia",
""
],
[
"Pedrini",
"Helio",
""
],
[
"Avila",
"Sandra",
""
]
] |
2508.10305
|
Kai Zhao
|
Ruoyu Li, Yafan Huang, Longtao Zhang, Zhuoxun Yang, Sheng Di, Jiajun Huang, Jinyang Liu, Jiannan Tian, Xin Liang, Guanpeng Li, Hanqi Guo, Franck Cappello, and Kai Zhao
|
GPZ: GPU-Accelerated Lossy Compressor for Particle Data
| null | null | null | null |
cs.DC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Particle-based simulations and point-cloud applications generate massive, irregular datasets that challenge storage, I/O, and real-time analytics. Traditional compression techniques struggle with irregular particle distributions and GPU architectural constraints, often resulting in limited throughput and suboptimal compression ratios. In this paper, we present GPZ, a high-performance, error-bounded lossy compressor designed specifically for large-scale particle data on modern GPUs. GPZ employs a novel four-stage parallel pipeline that synergistically balances high compression efficiency with the architectural demands of massively parallel hardware. We introduce a suite of targeted optimizations for computation, memory access, and GPU occupancy that enables GPZ to achieve near-hardware-limit throughput. We conduct an extensive evaluation on three distinct GPU architectures (workstation, data center, and edge) using six large-scale, real-world scientific datasets from five distinct domains. The results demonstrate that GPZ consistently and significantly outperforms five state-of-the-art GPU compressors, delivering up to 8x higher end-to-end throughput while simultaneously achieving superior compression ratios and data quality.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:23:32 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Li",
"Ruoyu",
""
],
[
"Huang",
"Yafan",
""
],
[
"Zhang",
"Longtao",
""
],
[
"Yang",
"Zhuoxun",
""
],
[
"Di",
"Sheng",
""
],
[
"Huang",
"Jiajun",
""
],
[
"Liu",
"Jinyang",
""
],
[
"Tian",
"Jiannan",
""
],
[
"Liang",
"Xin",
""
],
[
"Li",
"Guanpeng",
""
],
[
"Guo",
"Hanqi",
""
],
[
"Cappello",
"Franck",
""
],
[
"Zhao",
"Kai",
""
]
] |
2508.10306
|
Pawel Gajer
|
Pawel Gajer and Jacques Ravel
|
Intrinsic and Normal Mean Ricci Curvatures: A Bochner--Weitzenboeck Identity for Simple d-Vectors
|
12 pages, 1 figure
| null | null | null |
math.DG math.AP math.SP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We introduce two pointwise subspace averages of sectional curvature on a d-dimensional plane Pi in T_p M: (i) the intrinsic mean Ricci (the average of sectional curvatures of 2-planes contained in Pi); and (ii) the normal (mixed) mean Ricci (the average of sectional curvatures of 2-planes spanned by one vector in Pi and one in Pi^perp). Using Jacobi-field expansions, these means occur as the r^2/6 coefficients in the intrinsic (d-1)-sphere and normal (n-d-1)-sphere volume elements. A direct consequence is a Bochner--Weitzenboeck identity for simple d-vectors V (built from an orthonormal frame X_1,...,X_d with Pi = span{X_i}): the curvature term equals d(n-d) times the normal mean Ricci of Pi. This yields two immediate applications: (a) a Bochner vanishing criterion for harmonic simple d-vectors under a positive lower bound on the normal mean Ricci; and (b) a Lichnerowicz-type lower bound for the first eigenvalue of the Hodge Laplacian on simple d-eigenfields.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:25:04 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Gajer",
"Pawel",
""
],
[
"Ravel",
"Jacques",
""
]
] |
2508.10307
|
Zhaoming Kong
|
Zhaoming Kong, Jiahuan Zhang, Xiaowei Yang
|
Efficient Image Denoising Using Global and Local Circulant Representation
| null | null | null | null |
eess.IV cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
The advancement of imaging devices and countless image data generated everyday impose an increasingly high demand on efficient and effective image denoising. In this paper, we present a computationally simple denoising algorithm, termed Haar-tSVD, aiming to explore the nonlocal self-similarity prior and leverage the connection between principal component analysis (PCA) and the Haar transform under circulant representation. We show that global and local patch correlations can be effectively captured through a unified tensor-singular value decomposition (t-SVD) projection with the Haar transform. This results in a one-step, highly parallelizable filtering method that eliminates the need for learning local bases to represent image patches, striking a balance between denoising speed and performance. Furthermore, we introduce an adaptive noise estimation scheme based on a CNN estimator and eigenvalue analysis to enhance the robustness and adaptability of the proposed method. Experiments on different real-world denoising tasks validate the efficiency and effectiveness of Haar-tSVD for noise removal and detail preservation. Datasets, code and results are publicly available at https://github.com/ZhaomingKong/Haar-tSVD.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:25:59 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Kong",
"Zhaoming",
""
],
[
"Zhang",
"Jiahuan",
""
],
[
"Yang",
"Xiaowei",
""
]
] |
2508.10308
|
Sihang Zeng
|
Sihang Zeng, Kai Tian, Kaiyan Zhang, Yuru wang, Junqi Gao, Runze Liu, Sa Yang, Jingxuan Li, Xinwei Long, Jiaheng Ma, Biqing Qi, Bowen Zhou
|
ReviewRL: Towards Automated Scientific Review with RL
|
13 pages, 5 figures
| null | null | null |
cs.CL cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Peer review is essential for scientific progress but faces growing challenges due to increasing submission volumes and reviewer fatigue. Existing automated review approaches struggle with factual accuracy, rating consistency, and analytical depth, often generating superficial or generic feedback lacking the insights characteristic of high-quality human reviews. We introduce ReviewRL, a reinforcement learning framework for generating comprehensive and factually grounded scientific paper reviews. Our approach combines: (1) an ArXiv-MCP retrieval-augmented context generation pipeline that incorporates relevant scientific literature, (2) supervised fine-tuning that establishes foundational reviewing capabilities, and (3) a reinforcement learning procedure with a composite reward function that jointly enhances review quality and rating accuracy. Experiments on ICLR 2025 papers demonstrate that ReviewRL significantly outperforms existing methods across both rule-based metrics and model-based quality assessments. ReviewRL establishes a foundational framework for RL-driven automatic critique generation in scientific discovery, demonstrating promising potential for future development in this domain. The implementation of ReviewRL will be released at GitHub.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:26:13 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zeng",
"Sihang",
""
],
[
"Tian",
"Kai",
""
],
[
"Zhang",
"Kaiyan",
""
],
[
"wang",
"Yuru",
""
],
[
"Gao",
"Junqi",
""
],
[
"Liu",
"Runze",
""
],
[
"Yang",
"Sa",
""
],
[
"Li",
"Jingxuan",
""
],
[
"Long",
"Xinwei",
""
],
[
"Ma",
"Jiaheng",
""
],
[
"Qi",
"Biqing",
""
],
[
"Zhou",
"Bowen",
""
]
] |
2508.10309
|
Wenjie Zhao
|
Wenjie Zhao, Jia Li, Yunhui Guo
|
From Pixel to Mask: A Survey of Out-of-Distribution Segmentation
| null | null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Out-of-distribution (OoD) detection and segmentation have attracted growing attention as concerns about AI security rise. Conventional OoD detection methods identify the existence of OoD objects but lack spatial localization, limiting their usefulness in downstream tasks. OoD segmentation addresses this limitation by localizing anomalous objects at pixel-level granularity. This capability is crucial for safety-critical applications such as autonomous driving, where perception modules must not only detect but also precisely segment OoD objects, enabling targeted control actions and enhancing overall system robustness. In this survey, we group current OoD segmentation approaches into four categories: (i) test-time OoD segmentation, (ii) outlier exposure for supervised training, (iii) reconstruction-based methods, (iv) and approaches that leverage powerful models. We systematically review recent advances in OoD segmentation for autonomous-driving scenarios, identify emerging challenges, and discuss promising future research directions.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:26:56 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhao",
"Wenjie",
""
],
[
"Li",
"Jia",
""
],
[
"Guo",
"Yunhui",
""
]
] |
2508.10310
|
Kaixun Yang
|
Kaixun Yang, Yizhou Fan, Luzhen Tang, Mladen Rakovi\'c, Xinyu Li, Dragan Ga\v{s}evi\'c, Guanliang Chen
|
Beyond Self-Regulated Learning Processes: Unveiling Hidden Tactics in Generative AI-Assisted Writing
| null | null | null | null |
cs.HC cs.CY
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The integration of Generative AI (GenAI) into education is reshaping how students learn, making self-regulated learning (SRL) - the ability to plan, monitor, and adapt one's learning - more important than ever. To support learners in these new contexts, it is essential to understand how SRL unfolds during interaction with GenAI tools. Learning analytics offers powerful techniques for analyzing digital trace data to infer SRL behaviors. However, existing approaches often assume SRL processes are linear, segmented, and non-overlapping-assumptions that overlook the dynamic, recursive, and non-linear nature of real-world learning. We address this by conceptualizing SRL as a layered system: observable learning patterns reflect hidden tactics (short, purposeful action states), which combine into broader SRL strategies. Using Hidden Markov Models (HMMs), we analyzed trace data from higher education students engaged in GenAI-assisted academic writing. We identified three distinct groups of learners, each characterized by different SRL strategies. These groups showed significant differences in performance, indicating that students' use of different SRL strategies in GenAI-assisted writing led to varying task outcomes. Our findings advance the methodological toolkit for modeling SRL and inform the design of adaptive learning technologies that more effectively support learners in GenAI-enhanced educational environments.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:28:36 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Yang",
"Kaixun",
""
],
[
"Fan",
"Yizhou",
""
],
[
"Tang",
"Luzhen",
""
],
[
"Raković",
"Mladen",
""
],
[
"Li",
"Xinyu",
""
],
[
"Gašević",
"Dragan",
""
],
[
"Chen",
"Guanliang",
""
]
] |
2508.10311
|
Xuan Li
|
Xuan Li, Jialiang Dong, Raymond Wong
|
From Surface to Semantics: Semantic Structure Parsing for Table-Centric Document Analysis
|
8 pages, 5 figures, 28th European Conference on Artificial Intelligence (ECAI-2025)
| null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Documents are core carriers of information and knowl-edge, with broad applications in finance, healthcare, and scientific research. Tables, as the main medium for structured data, encapsulate key information and are among the most critical document components. Existing studies largely focus on surface-level tasks such as layout analysis, table detection, and data extraction, lacking deep semantic parsing of tables and their contextual associations. This limits advanced tasks like cross-paragraph data interpretation and context-consistent analysis. To address this, we propose DOTABLER, a table-centric semantic document parsing framework designed to uncover deep semantic links between tables and their context. DOTABLER leverages a custom dataset and domain-specific fine-tuning of pre-trained models, integrating a complete parsing pipeline to identify context segments semantically tied to tables. Built on this semantic understanding, DOTABLER implements two core functionalities: table-centric document structure parsing and domain-specific table retrieval, delivering comprehensive table-anchored semantic analysis and precise extraction of semantically relevant tables. Evaluated on nearly 4,000 pages with over 1,000 tables from real-world PDFs, DOTABLER achieves over 90% Precision and F1 scores, demonstrating superior performance in table-context semantic analysis and deep document parsing compared to advanced models such as GPT-4o.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:29:51 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Li",
"Xuan",
""
],
[
"Dong",
"Jialiang",
""
],
[
"Wong",
"Raymond",
""
]
] |
2508.10312
|
Minhao Wang
|
Minhao Wang, Yunhang He, Cong Xu, Zhangchi Zhu, Wei Zhang
|
Beyond Semantic Understanding: Preserving Collaborative Frequency Components in LLM-based Recommendation
|
12 pages, 8 figures
| null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recommender systems in concert with Large Language Models (LLMs) present promising avenues for generating semantically-informed recommendations. However, LLM-based recommenders exhibit a tendency to overemphasize semantic correlations within users' interaction history. When taking pretrained collaborative ID embeddings as input, LLM-based recommenders progressively weaken the inherent collaborative signals as the embeddings propagate through LLM backbones layer by layer, as opposed to traditional Transformer-based sequential models in which collaborative signals are typically preserved or even enhanced for state-of-the-art performance. To address this limitation, we introduce FreLLM4Rec, an approach designed to balance semantic and collaborative information from a spectral perspective. Item embeddings that incorporate both semantic and collaborative information are first purified using a Global Graph Low-Pass Filter (G-LPF) to preliminarily remove irrelevant high-frequency noise. Temporal Frequency Modulation (TFM) then actively preserves collaborative signal layer by layer. Note that the collaborative preservation capability of TFM is theoretically guaranteed by establishing a connection between the optimal but hard-to-implement local graph fourier filters and the suboptimal yet computationally efficient frequency-domain filters. Extensive experiments on four benchmark datasets demonstrate that FreLLM4Rec successfully mitigates collaborative signal attenuation and achieves competitive performance, with improvements of up to 8.00\% in NDCG@10 over the best baseline. Our findings provide insights into how LLMs process collaborative information and offer a principled approach for improving LLM-based recommendation systems.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:33:02 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Wang",
"Minhao",
""
],
[
"He",
"Yunhang",
""
],
[
"Xu",
"Cong",
""
],
[
"Zhu",
"Zhangchi",
""
],
[
"Zhang",
"Wei",
""
]
] |
2508.10313
|
Jixiang Chen
|
Jixiang Chen, Yiqun Lin, Yi Qin, Hualiang Wang, Xiaomeng Li
|
Cross-view Generalized Diffusion Model for Sparse-view CT Reconstruction
|
MICCAI 2025 Spotlight
| null | null | null |
eess.IV
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Sparse-view computed tomography (CT) reduces radiation exposure by subsampling projection views, but conventional reconstruction methods produce severe streak artifacts with undersampled data. While deep-learning-based methods enable single-step artifact suppression, they often produce over-smoothed results under significant sparsity. Though diffusion models improve reconstruction via iterative refinement and generative priors, they require hundreds of sampling steps and struggle with stability in highly sparse regimes. To tackle these concerns, we present the Cross-view Generalized Diffusion Model (CvG-Diff), which reformulates sparse-view CT reconstruction as a generalized diffusion process. Unlike existing diffusion approaches that rely on stochastic Gaussian degradation, CvG-Diff explicitly models image-domain artifacts caused by angular subsampling as a deterministic degradation operator, leveraging correlations across sparse-view CT at different sample rates. To address the inherent artifact propagation and inefficiency of sequential sampling in generalized diffusion model, we introduce two innovations: Error-Propagating Composite Training (EPCT), which facilitates identifying error-prone regions and suppresses propagated artifacts, and Semantic-Prioritized Dual-Phase Sampling (SPDPS), an adaptive strategy that prioritizes semantic correctness before detail refinement. Together, these innovations enable CvG-Diff to achieve high-quality reconstructions with minimal iterations, achieving 38.34 dB PSNR and 0.9518 SSIM for 18-view CT using only \textbf{10} steps on AAPM-LDCT dataset. Extensive experiments demonstrate the superiority of CvG-Diff over state-of-the-art sparse-view CT reconstruction methods. The code is available at https://github.com/xmed-lab/CvG-Diff.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:33:50 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Chen",
"Jixiang",
""
],
[
"Lin",
"Yiqun",
""
],
[
"Qin",
"Yi",
""
],
[
"Wang",
"Hualiang",
""
],
[
"Li",
"Xiaomeng",
""
]
] |
2508.10314
|
Kensuke Yoshizawa
|
Tatsuya Miura, Kensuke Yoshizawa
|
Stability of flat-core pinned p-elasticae
|
12 pages, 3 figures
| null | null | null |
math.AP math.DG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We classify the stability of flat-core $p$-elasticae in $\mathbf{R}^d$ subject to the pinned boundary condition. Together with previous work, this completes the classification of stable pinned $p$-elasticae in $\mathbf{R}^d$ for all $p\in(1,\infty)$ and $d\geq2$.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:36:15 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Miura",
"Tatsuya",
""
],
[
"Yoshizawa",
"Kensuke",
""
]
] |
2508.10315
|
Dongjue Wang
|
Keke Gai, Dongjue Wang, Jing Yu, Liehuang Zhu, Qi Wu
|
A Vision-Language Pre-training Model-Guided Approach for Mitigating Backdoor Attacks in Federated Learning
| null | null | null | null |
cs.LG cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Existing backdoor defense methods in Federated Learning (FL) rely on the assumption of homogeneous client data distributions or the availability of a clean serve dataset, which limits the practicality and effectiveness. Defending against backdoor attacks under heterogeneous client data distributions while preserving model performance remains a significant challenge. In this paper, we propose a FL backdoor defense framework named CLIP-Fed, which leverages the zero-shot learning capabilities of vision-language pre-training models. By integrating both pre-aggregation and post-aggregation defense strategies, CLIP-Fed overcomes the limitations of Non-IID imposed on defense effectiveness. To address privacy concerns and enhance the coverage of the dataset against diverse triggers, we construct and augment the server dataset using the multimodal large language model and frequency analysis without any client samples. To address class prototype deviations caused by backdoor samples and eliminate the correlation between trigger patterns and target labels, CLIP-Fed aligns the knowledge of the global model and CLIP on the augmented dataset using prototype contrastive loss and Kullback-Leibler divergence. Extensive experiments on representative datasets validate the effectiveness of CLIP-Fed. Compared to state-of-the-art methods, CLIP-Fed achieves an average reduction in ASR, i.e., 2.03\% on CIFAR-10 and 1.35\% on CIFAR-10-LT, while improving average MA by 7.92\% and 0.48\%, respectively.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:39:54 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Gai",
"Keke",
""
],
[
"Wang",
"Dongjue",
""
],
[
"Yu",
"Jing",
""
],
[
"Zhu",
"Liehuang",
""
],
[
"Wu",
"Qi",
""
]
] |
2508.10316
|
Yuanzhi Liang
|
Yuanzhi Liang, Yijie Fang, Rui Li, Ziqi Ni, Ruijie Su, Chi Zhang, Xuelong Li
|
Integrating Reinforcement Learning with Visual Generative Models: Foundations and Advances
|
Ongoing work
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Generative models have made significant progress in synthesizing visual content, including images, videos, and 3D/4D structures. However, they are typically trained with surrogate objectives such as likelihood or reconstruction loss, which often misalign with perceptual quality, semantic accuracy, or physical realism. Reinforcement learning (RL) offers a principled framework for optimizing non-differentiable, preference-driven, and temporally structured objectives. Recent advances demonstrate its effectiveness in enhancing controllability, consistency, and human alignment across generative tasks. This survey provides a systematic overview of RL-based methods for visual content generation. We review the evolution of RL from classical control to its role as a general-purpose optimization tool, and examine its integration into image, video, and 3D/4D generation. Across these domains, RL serves not only as a fine-tuning mechanism but also as a structural component for aligning generation with complex, high-level goals. We conclude with open challenges and future research directions at the intersection of RL and generative modeling.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:44:03 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Liang",
"Yuanzhi",
""
],
[
"Fang",
"Yijie",
""
],
[
"Li",
"Rui",
""
],
[
"Ni",
"Ziqi",
""
],
[
"Su",
"Ruijie",
""
],
[
"Zhang",
"Chi",
""
],
[
"Li",
"Xuelong",
""
]
] |
2508.10317
|
Xiaoming Chen
|
Yichao Xu, Xiaoming Chen, Ming Ying, Zhaoyang Zhang
|
Integrated Communication and Remote Sensing in LEO Satellite Systems: Protocol, Architecture and Prototype
| null |
IEEE Transactions on Wireless Communications, 2025
| null | null |
cs.IT eess.SP math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we explore the integration of communication and synthetic aperture radar (SAR)-based remote sensing in low Earth orbit (LEO) satellite systems to provide real-time SAR imaging and information transmission. Considering the high-mobility characteristics of satellite channels and limited processing capabilities of satellite payloads, we propose an integrated communication and remote sensing architecture based on an orthogonal delay-Doppler division multiplexing (ODDM) signal waveform. Both communication and SAR imaging functionalities are achieved with an integrated transceiver onboard the LEO satellite, utilizing the same waveform and radio frequency (RF) front-end. Based on such an architecture, we propose a transmission protocol compatible with the 5G NR standard using downlink pilots for joint channel estimation and SAR imaging. Furthermore, we design a unified signal processing framework for the integrated satellite receiver to simultaneously achieve high-performance channel sensing, low-complexity channel equalization and interference-free SAR imaging. Finally, the performance of the proposed integrated system is demonstrated through comprehensive analysis and extensive simulations in the sub-6 GHz band. Moreover, a software-defined radio (SDR) prototype is presented to validate its effectiveness for real-time SAR imaging and information transmission in satellite direct-connect user equipment (UE) scenarios within the millimeter-wave (mmWave) band.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:44:45 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Xu",
"Yichao",
""
],
[
"Chen",
"Xiaoming",
""
],
[
"Ying",
"Ming",
""
],
[
"Zhang",
"Zhaoyang",
""
]
] |
2508.10318
|
Huangbin Liang
|
Huangbin Liang and Beatriz Moya and Francisco Chinesta and Eleni Chatzi
|
Quantifying the Value of Seismic Structural Health Monitoring for post-earthquake recovery of electric power system in terms of resilience enhancement
|
21 pages. 14 figures
| null | null | null |
eess.SY cs.SY
|
http://creativecommons.org/licenses/by/4.0/
|
Post-earthquake recovery of electric power networks (EPNs) is critical to community resilience. Traditional recovery processes often rely on prolonged and imprecise manual inspections for damage diagnosis, leading to suboptimal repair prioritization and extended service disruptions. Seismic Structural Health Monitoring (SSHM) offers the potential to expedite recovery by enabling more accurate and timely damage assessment. However, SSHM deployment incurs costs, and its system-level resilience benefit remains underexplored. This study proposes a probabilistic simulation framework to quantify the value of SSHM for enhancing EPN resilience. The framework includes seismic damage modeling based on network configuration, hazard intensity, fragility functions, and damage-functionality mappings, combined with recovery simulations incorporating resource constraints, repair and transfer durations. System functionality is evaluated using graph-based island detection and optimal power flow analysis. Resilience is quantified via the Lack of Resilience (LoR) metric derived from the functionality restoration curve. SSHM is incorporated by altering the quality of damage information used in repair scheduling. Different monitoring scenarios (e.g., no-SSHM baseline, partial SSHM, full SSHM with various accuracies) are modeled using confusion matrices to simulate damage misclassification. Results show that improved damage awareness via SSHM significantly accelerates recovery and reduces LoR by up to 21%. This work supports evidence-based decisions for SSHM deployment in critical infrastructure.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:46:24 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Liang",
"Huangbin",
""
],
[
"Moya",
"Beatriz",
""
],
[
"Chinesta",
"Francisco",
""
],
[
"Chatzi",
"Eleni",
""
]
] |
2508.10319
|
Jiarui Sun
|
Jiarui Sun, Jun Zhang, Li Cui, Alessandro Sonnenfeld and Xin Wang
|
Precision Measurement of Large Shear Signals
|
7 pages, 3 figures, submitted to ApJL
| null | null | null |
astro-ph.CO
|
http://creativecommons.org/licenses/by/4.0/
|
So far, estimators of galaxy shape distortions are only carefully studied perturbatively in the case of small shear signals, mainly for weak lensing sciences. However, in the neighborhood of massive foreground clusters, a large number of background galaxies can be significantly distorted. The measurement of such large shear signals could be quite nontrivial under general observing conditions, i.e., in the presence of the point spread function (PSF) and noise. In this work, we propose a non-perturbative method to exactly recover large shear signals ($\gtrsim 0.5$) under general conditions. We test the method on simulated galaxy images, and find that it is accurate down to the very faint end.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:48:58 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Sun",
"Jiarui",
""
],
[
"Zhang",
"Jun",
""
],
[
"Cui",
"Li",
""
],
[
"Sonnenfeld",
"Alessandro",
""
],
[
"Wang",
"Xin",
""
]
] |
2508.10320
|
Aaditya Chandrasekhar
|
Aaditya Chandrasekhar, Stefan Knapik, Deepak Sharma, John Reidy, Ian McCue, Jian Cao, Wei Chen
|
TOBACO: Topology Optimization via Band-limited Coordinate Networks for Compositionally Graded Alloys
|
Submitted to Structural and Multidisciplinary Optimization
| null | null | null |
cs.CE cs.NA math.NA
|
http://creativecommons.org/licenses/by/4.0/
|
Compositionally Graded Alloys (CGAs) offer unprecedented design flexibility by enabling spatial variations in composition; tailoring material properties to local loading conditions. This flexibility leads to components that are stronger, lighter, and more cost-effective than traditional monolithic counterparts. The fabrication of CGAs have become increasingly feasible owing to recent advancements in additive manufacturing (AM), particularly in multi-material printing and improved precision in material deposition. However, AM of CGAs requires imposition of manufacturing constraints; in particular limits on the maximum spatial gradation of composition.
This paper introduces a topology optimization (TO) based framework for designing optimized CGA components with controlled compositional gradation. In particular, we represent the constrained composition distribution using a band-limited coordinate neural network. By regulating the network's bandwidth, we ensure implicit compliance with gradation limits, eliminating the need for explicit constraints. The proposed approach also benefits from the inherent advantages of TO using coordinate networks, including mesh independence, high-resolution design extraction, and end-to-end differentiability. The effectiveness of our framework is demonstrated through various elastic and thermo-elastic TO examples.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:49:09 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Chandrasekhar",
"Aaditya",
""
],
[
"Knapik",
"Stefan",
""
],
[
"Sharma",
"Deepak",
""
],
[
"Reidy",
"John",
""
],
[
"McCue",
"Ian",
""
],
[
"Cao",
"Jian",
""
],
[
"Chen",
"Wei",
""
]
] |
2508.10321
|
James Tian
|
James Tian
|
Random Operator-Valued Kernels and Moment Dilations
| null | null | null | null |
math.FA
|
http://creativecommons.org/licenses/by/4.0/
|
This paper studies random operator-valued positive definite (p.d.) kernels and their connection to moment dilations. A class of random p.d. kernels is introduced in which the positivity requirement is imposed only in expectation, extending the deterministic case. Basic properties are established, including convexity, sampling stability, and the relationship between mean kernels and deterministic ones. A Gaussian factorization result is proved under a Hilbert-Schmidt/trace condition, clarifying the difference between pathwise and mean-square positive definiteness and linking the construction to radonification in the non-trace-class setting. Moment dilation is then formulated for random operators, extending classical dilation theory to preserve equality of mixed moments in expectation. The resulting dilation triples admit interpretation in terms of unitary dynamics, connecting statistical moment structure with operator-theoretic realization.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:50:02 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Tian",
"James",
""
]
] |
2508.10322
|
Chuqi Chen
|
Qixuan Zhou, Chuqi Chen, Tao Luo, Yang Xiang
|
SSBE-PINN: A Sobolev Boundary Scheme Boosting Stability and Accuracy in Elliptic/Parabolic PDE Learning
| null | null | null | null |
math.NA cs.NA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Physics-Informed Neural Networks (PINNs) have emerged as a powerful framework for solving partial differential equations (PDEs), yet they often fail to achieve accurate convergence in the H1 norm, especially in the presence of boundary approximation errors. In this work, we propose a novel method called Sobolev-Stable Boundary Enforcement (SSBE), which redefines the boundary loss using Sobolev norms to incorporate boundary regularity directly into the training process. We provide rigorous theoretical analysis demonstrating that SSBE ensures bounded H1 error via a stability guarantee and derive generalization bounds that characterize its robustness under finite-sample regimes. Extensive numerical experiments on linear and nonlinear PDEs, including Poisson, heat, and elliptic problems, show that SSBE consistently outperforms standard PINNs in terms of both relative L2 and H1 errors, even in high-dimensional settings. The proposed approach offers a principled and practical solution for improving gradient fidelity and overall solution accuracy in neural network based PDE solvers.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:51:44 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhou",
"Qixuan",
""
],
[
"Chen",
"Chuqi",
""
],
[
"Luo",
"Tao",
""
],
[
"Xiang",
"Yang",
""
]
] |
2508.10323
|
Ryo Horiuchi
|
Ryo Horiuchi
|
On the Witt vectors in the Lawvere quantale
|
14 pages
| null | null | null |
math.CT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this note, we investigate the Witt vectors in the min-plus algebra of extended non-negative real numbers, and consider categories enriched over them viewed as a monoidal category.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:59:32 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Horiuchi",
"Ryo",
""
]
] |
2508.10324
|
Qi Liu
|
Jing Zhang, Qi Liu, Yongmo Hu, Linlin Fu, Yuxin Wang, Jinyu Xia, John Michael Rassias, Choonkil Park, Yongjin Li
|
Hyers-Ulam-Rassias stability of functional equations with parameters
| null | null | null | null |
math.FA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This paper explores the Hyers-Ulam stability of generalized Jensen additive and quadratic functional equations in \(\beta\)-homogeneous \(F\)-space, showing that approximately satisfying mappings have a unique exact approximating counterpart within a specific bound.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 03:59:33 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Jing",
""
],
[
"Liu",
"Qi",
""
],
[
"Hu",
"Yongmo",
""
],
[
"Fu",
"Linlin",
""
],
[
"Wang",
"Yuxin",
""
],
[
"Xia",
"Jinyu",
""
],
[
"Rassias",
"John Michael",
""
],
[
"Park",
"Choonkil",
""
],
[
"Li",
"Yongjin",
""
]
] |
2508.10325
|
Masashi Hazumi
|
Masashi Hazumi
|
Quantum measurement systems and applications to particle physics and cosmology
|
25 pages, 10 figures, contribution to the Vienna Conference on Instrumentation (VCI) 2025, 17-21 February 2025
|
Nucl.Instrum.Meth.A 1081 (2026) 170847
|
10.1016/j.nima.2025.170847
| null |
physics.ins-det astro-ph.IM hep-ex quant-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
There are new detector proposals and R&D that utilize quantum enhancements not previously adopted. Examples include superconducting quantum sensors, atom interferometry, and quantum spin sensors. They are mainly motivated by industrial applications toward quantum computing, secure quantum communication systems, and high-sensitivity sensors. Given the excellent potential of the new quantum measurement systems, there are also new proposals to apply them in particle physics and cosmology. In this review, I survey currently available and emerging quantum technologies and their applications. I then discuss future directions and new proposals for particle physics and cosmology.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:01:17 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Hazumi",
"Masashi",
""
]
] |
2508.10326
|
Nathan Long
|
Nathan K. Long, Ziqing Wang, Benjamin P. Dix-Matthews, Alex Frost, John Wallis, Kenneth J. Grant, Robert Malaney
|
Quantum Wavefront Correction via Machine Learning for Satellite-to-Earth CV-QKD
| null | null | null | null |
quant-ph eess.SP physics.optics
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
State-of-the-art free-space continuous-variable quantum key distribution (CV-QKD) protocols use phase reference pulses to modulate the wavefront of a real local oscillator at the receiver, thereby compensating for wavefront distortions caused by atmospheric turbulence. It is normally assumed that the wavefront distortion in these phase reference pulses is identical to the wavefront distortion in the quantum signals, which are multiplexed during transmission. However, in many real-world deployments, there can exist a relative wavefront error (WFE) between the reference pulses and quantum signals, which, among other deleterious effects, can severely limit secure key transfer in satellite-to-Earth CV-QKD. In this work, we introduce novel machine learning-based wavefront correction algorithms, which utilize multi-plane light conversion for decomposition of the reference pulses and quantum signals into the Hermite-Gaussian (HG) basis, then estimate the difference in HG mode phase measurements, effectively eliminating this problem. Through detailed simulations of the Earth-satellite channel, we demonstrate that our new algorithm can rapidly identify and compensate for any relative WFEs that may exist, whilst causing no harm when WFEs are similar across both the reference pulses and quantum signals. We quantify the gains available in our algorithm in terms of the CV-QKD secure key rate. We show channels where positive secure key rates are obtained using our algorithms, while information loss without wavefront correction would result in null key rates.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:03:07 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Long",
"Nathan K.",
""
],
[
"Wang",
"Ziqing",
""
],
[
"Dix-Matthews",
"Benjamin P.",
""
],
[
"Frost",
"Alex",
""
],
[
"Wallis",
"John",
""
],
[
"Grant",
"Kenneth J.",
""
],
[
"Malaney",
"Robert",
""
]
] |
2508.10327
|
Shuangquan Zhangsq
|
Haoyang Hu, Xun Huang, Chenyu Wu, Shiwen Liu, Zhichao Lian, Shuangquan Zhang
|
BERTector: Intrusion Detection Based on Joint-Dataset Learning
| null | null | null | null |
cs.CR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Intrusion detection systems (IDS) are facing challenges in generalization and robustness due to the heterogeneity of network traffic and the diversity of attack patterns. To address this issue, we propose a new joint-dataset training paradigm for IDS and propose a scalable BERTector framework based on BERT. BERTector integrates three key components: NSS-Tokenizer for traffic-aware semantic tokenization, supervised fine-tuning with a hybrid dataset, and low-rank adaptation (LoRA) for efficient training. Extensive experiments show that BERTector achieves state-of-the-art detection accuracy, strong cross-dataset generalization capabilities, and excellent robustness to adversarial perturbations. This work establishes a unified and efficient solution for modern IDS in complex and dynamic network environments.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:05:01 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Hu",
"Haoyang",
""
],
[
"Huang",
"Xun",
""
],
[
"Wu",
"Chenyu",
""
],
[
"Liu",
"Shiwen",
""
],
[
"Lian",
"Zhichao",
""
],
[
"Zhang",
"Shuangquan",
""
]
] |
2508.10328
|
Kazem Azizi
|
M. Shekari Tousi, K. Azizi
|
Doubly heavy spin-$\frac {3}{2} $ baryons spectrum in the ground and excited states
|
16 Pages, 4 Figures and 9 Tables
| null | null | null |
hep-ph hep-ex hep-lat
|
http://creativecommons.org/licenses/by/4.0/
|
This study employs QCD sum rules to predict the masses and residues of spin-$\frac {3}{2} $ doubly heavy baryons including two heavy quarks (c and/or b) and one light quark, specifically focusing on $ \Xi_{cc}^*$, $ \Xi_{bc}^*$, $ \Xi_{bb}^*$, $ \Omega_{cc}^*$, $ \Omega_{bc}^*$ and $ \Omega_{bb}^*$. Our study provides results for the ground state ($1S$), first orbital excitation ($1P$), and the first radial excitation ($2S$), within a consistent theoretical framework. In addition to mass spectra, we provide residue calculations as well. The calculated residues are essential for estimating the decay widths and branching ratios of these baryons at different decay channels. Our analysis incorporates nonperturbative QCD effects through operators up to dimension ten, leading to improved precision in the mass and residue calculations. These predictions offer crucial guidance for ongoing and future experimental searches, particularly in light of the current lack of empirical data for the ground and excited states, and provide a basis for comparison with future experimental data.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:05:42 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Tousi",
"M. Shekari",
""
],
[
"Azizi",
"K.",
""
]
] |
2508.10329
|
Jialu Nie
|
Jialu Nie, Peng Wei, Zihuang Cao, Yibo Yan, Chao Liu, Hao Tian, Xin Zhang and Haijun Tian
|
Toward High-Precision Astrometry with CSST Using Multi-Gaussian Fitting of PSF
| null | null | null | null |
astro-ph.IM astro-ph.GA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The Chinese Space Station Survey Telescope (CSST) presents significant potential for high-precision astrometry. In this study, we show that the point spread function (PSF) modeled by the discrete PSF with Multi-Gaussian function can effectively enhance the astrometric accuracy. We determine that the PSF profile can be accurately modeled by three Gaussians, which takes advantage of reduced computational complexity in PSF convolution. In sparse star fields, the lowest centering accuracy we obtain after aberration correction can be below 1 mas. We find that the proper motion errors remain below 1.0 mas/yr for point sources with five observations and approximately 0.8 mas/yr for seven observations with a time baseline of around 3.5 years. We finally demonstrate that the precision of our position measurements for stars fainter than 21 mag in the simulated CSST crowded field is better than the results from both SExtractor and DOLPHOT.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:06:30 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Nie",
"Jialu",
""
],
[
"Wei",
"Peng",
""
],
[
"Cao",
"Zihuang",
""
],
[
"Yan",
"Yibo",
""
],
[
"Liu",
"Chao",
""
],
[
"Tian",
"Hao",
""
],
[
"Zhang",
"Xin",
""
],
[
"Tian",
"Haijun",
""
]
] |
2508.10330
|
Christopher Lee
|
C. P. Lee, N. D. R. Bhat, B. W. Meyers, S. J. McSweeney, W. van Straten, C. M. Tan, M. Xue, N. A. Swainston, S. M. Ord, G. Sleap, S. E. Tremblay, A. Williams
|
The Southern-sky MWA Rapid Two-metre (SMART) pulsar survey -- III. A census of millisecond pulsars at 154 MHz
|
22 pages, 10 figures, 2 tables, accepted for publication in PASA
| null | null | null |
astro-ph.HE
|
http://creativecommons.org/licenses/by/4.0/
|
Observations of millisecond pulsars (MSPs) at low radio frequencies play an important role in understanding the Galactic pulsar population and characterising both their emission properties and the effects of the ionised interstellar medium on the received signals. To date, only a relatively small fraction of the known MSP population has been detected at frequencies below 300 MHz, and nearly all previous MSP studies at these frequencies have been conducted with northern telescopes. We present a census of MSPs in the SMART pulsar survey, covering declinations south of +30 deg at a centre frequency of 154 MHz. We detected 40 MSPs, with 11 being the first published detections below 300 MHz. For each detection, we provide coherently-dedispersed full-polarimetric integrated pulse profiles and mean flux densities. We measured significant Faraday rotation measures (RMs) for 25 MSPs, and identified apparent phase-dependent RM variations for three MSPs. Comparison with published profiles at other frequencies supports previous studies suggesting that the pulse component separations of MSPs vary negligibly over a wide frequency range due to their compact magnetospheres. We observe that integrated pulse profiles tend to be more polarised at low frequencies, consistent with depolarisation due to superposed orthogonal polarisation modes. The results of this census will be a valuable resource for planning future MSP monitoring projects at low frequencies, and will also help to improve survey simulations to forecast the detectable MSP population with the SKA-Low.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:08:50 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Lee",
"C. P.",
""
],
[
"Bhat",
"N. D. R.",
""
],
[
"Meyers",
"B. W.",
""
],
[
"McSweeney",
"S. J.",
""
],
[
"van Straten",
"W.",
""
],
[
"Tan",
"C. M.",
""
],
[
"Xue",
"M.",
""
],
[
"Swainston",
"N. A.",
""
],
[
"Ord",
"S. M.",
""
],
[
"Sleap",
"G.",
""
],
[
"Tremblay",
"S. E.",
""
],
[
"Williams",
"A.",
""
]
] |
2508.10331
|
Zhenkang Peng
|
Zhenkang Peng, Chengzhang Li, Ying Rong, Renyu (Philip) Zhang
|
Synthesizing Evidence: Data-Pooling as a Tool for Treatment Selection in Online Experiments
| null | null | null | null |
stat.ME
|
http://creativecommons.org/licenses/by/4.0/
|
Randomized experiments are the gold standard for causal inference but face significant challenges in business applications, including limited traffic allocation, the need for heterogeneous treatment effect estimation, and the complexity of managing overlapping experiments. These factors lead to high variability in treatment effect estimates, making data-driven policy roll out difficult. To address these issues, we introduce the data pooling treatment roll-out (DTR) framework, which enhances policy roll-out by pooling data across experiments rather than focusing narrowly on individual ones. DTR can effectively accommodate both overlapping and non-overlapping traffic scenarios, regardless of linear or nonlinear model specifications. We demonstrate the framework's robustness through a three-pronged validation: (a) theoretical analysis shows that DTR surpasses the traditional difference-in-mean and ordinary least squares methods under non-overlapping experiments, particularly when the number of experiments is large; (b) synthetic simulations confirm its adaptability in complex scenarios with overlapping traffic, rich covariates and nonlinear specifications; and (c) empirical applications to two experimental datasets from real world platforms, demonstrating its effectiveness in guiding customized policy roll-outs for subgroups within a single experiment, as well as in coordinating policy deployments across multiple experiments with overlapping scenarios. By reducing estimation variability to improve decision-making effectiveness, DTR provides a scalable, practical solution for online platforms to better leverage their experimental data in today's increasingly complex business environments.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:11:09 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Peng",
"Zhenkang",
"",
"Philip"
],
[
"Li",
"Chengzhang",
"",
"Philip"
],
[
"Rong",
"Ying",
"",
"Philip"
],
[
"Renyu",
"",
"",
"Philip"
],
[
"Zhang",
"",
""
]
] |
2508.10332
|
Sudarsana Kadiri
|
Abhijit Sinha, Harishankar Kumar, Mohit Joshi, Hemant Kumar Kathania, Shrikanth Narayanan, Sudarsana Reddy Kadiri
|
Layer-Wise Analysis of Self-Supervised Representations for Age and Gender Classification in Children's Speech
|
Accepted at Workshop on Child Computer Interaction (WOCCI 2025)
| null | null | null |
eess.AS cs.AI cs.HC cs.LG cs.SD
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Children's speech presents challenges for age and gender classification due to high variability in pitch, articulation, and developmental traits. While self-supervised learning (SSL) models perform well on adult speech tasks, their ability to encode speaker traits in children remains underexplored. This paper presents a detailed layer-wise analysis of four Wav2Vec2 variants using the PFSTAR and CMU Kids datasets. Results show that early layers (1-7) capture speaker-specific cues more effectively than deeper layers, which increasingly focus on linguistic information. Applying PCA further improves classification, reducing redundancy and highlighting the most informative components. The Wav2Vec2-large-lv60 model achieves 97.14% (age) and 98.20% (gender) on CMU Kids; base-100h and large-lv60 models reach 86.05% and 95.00% on PFSTAR. These results reveal how speaker traits are structured across SSL model depth and support more targeted, adaptive strategies for child-aware speech interfaces.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:11:44 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Sinha",
"Abhijit",
""
],
[
"Kumar",
"Harishankar",
""
],
[
"Joshi",
"Mohit",
""
],
[
"Kathania",
"Hemant Kumar",
""
],
[
"Narayanan",
"Shrikanth",
""
],
[
"Kadiri",
"Sudarsana Reddy",
""
]
] |
2508.10333
|
Wenxuan Song
|
Wenxuan Song, Ziyang Zhou, Han Zhao, Jiayi Chen, Pengxiang Ding, Haodong Yan, Yuxin Huang, Feilong Tang, Donglin Wang, Haoang Li
|
ReconVLA: Reconstructive Vision-Language-Action Model as Effective Robot Perceiver
| null | null | null | null |
cs.RO cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent advances in Vision-Language-Action (VLA) models have enabled robotic agents to integrate multimodal understanding with action execution. However, our empirical analysis reveals that current VLAs struggle to allocate visual attention to target regions. Instead, visual attention is always dispersed. To guide the visual attention grounding on the correct target, we propose ReconVLA, a reconstructive VLA model with an implicit grounding paradigm. Conditioned on the model's visual outputs, a diffusion transformer aims to reconstruct the gaze region of the image, which corresponds to the target manipulated objects. This process prompts the VLA model to learn fine-grained representations and accurately allocate visual attention, thus effectively leveraging task-specific visual information and conducting precise manipulation. Moreover, we curate a large-scale pretraining dataset comprising over 100k trajectories and 2 million data samples from open-source robotic datasets, further boosting the model's generalization in visual reconstruction. Extensive experiments in simulation and the real world demonstrate the superiority of our implicit grounding method, showcasing its capabilities of precise manipulation and generalization. Our project page is https://zionchow.github.io/ReconVLA/.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:20:19 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Song",
"Wenxuan",
""
],
[
"Zhou",
"Ziyang",
""
],
[
"Zhao",
"Han",
""
],
[
"Chen",
"Jiayi",
""
],
[
"Ding",
"Pengxiang",
""
],
[
"Yan",
"Haodong",
""
],
[
"Huang",
"Yuxin",
""
],
[
"Tang",
"Feilong",
""
],
[
"Wang",
"Donglin",
""
],
[
"Li",
"Haoang",
""
]
] |
2508.10334
|
Karl Sandeman
|
S. Vela, J. Ribas-Ari\~no, S.P. Vallone, A.M. dos Santos, M.A. Halcrow, K.G. Sandeman
|
Atomistic Description of Spin Crossover Under Pressure and its Giant Barocaloric Effect
|
Main text: 8 pages, 7 figures (main text); Supporting information: 4 pages, 4 figures
| null | null | null |
cond-mat.mtrl-sci cond-mat.str-el
|
http://creativecommons.org/licenses/by/4.0/
|
The pressure-dependent evolution of the spin crossover (SCO) transition has garnered significant interest due to its connection to the giant barocaloric effect (BCE) near room temperature. Pressure alters both the molecular and solid-state structures of SCO materials, affecting the relative stability of low- and high-spin states and, consequently, the transition temperature ($T_{1/2}$). Crucially, the shape of the $T_{1/2}$ vs. pressure curve dictates the magnitude of the BCE, making its accurate characterization essential for identifying high-performance materials. In this work, we investigate the nonlinear $T_{1/2}$ vs. pressure behavior of the prototypical SCO complex [FeL$_2$][BF$_4$]$_2$ [L = 2,6-di(pyrazol-1-yl)pyridine] using solid-state PBE+U computations. Our results unveil the mechanisms by which pressure influences its SCO transition, including the onset of a phase transition, as well as the key role of low-frequency phonons in the BCE. Furthermore, we establish a computational protocol for accurately modeling the BCE in SCO crystals, providing a powerful tool for the rapid and efficient discovery of new materials with enhanced barocaloric performance.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:25:07 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Vela",
"S.",
""
],
[
"Ribas-Ariño",
"J.",
""
],
[
"Vallone",
"S. P.",
""
],
[
"Santos",
"A. M. dos",
""
],
[
"Halcrow",
"M. A.",
""
],
[
"Sandeman",
"K. G.",
""
]
] |
2508.10335
|
Subhojoy Gupta
|
Subhojoy Gupta and Gobinda Sau
|
Harmonic maps and framed $\mathrm{PSL}_2(\mathbb{C})$-representations
|
27 pages, 3 figures
| null | null | null |
math.DG math.GT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We show that given an element $X$ of the enhanced Teichm\"{u}ller space $\mathcal{T}^\pm(\mathbb{S}, \mathbb{M})$ and a type-preserving framed $\mathrm{PSL}_2(\mathbb{C})$-representation $\hat{\rho} = (\rho,\beta)$, there is a $\rho$-equivariant harmonic map $f:\mathbb{H}^2 \to \mathbb{H}^3$ that is asymptotic to the framing $\beta$. Here, the domain is the universal cover of the punctured Riemann surface obtained from a conformal completion of $X$. Moreover, such a harmonic map is unique if one prescribes, in addition, the principal part of the Hopf differential at each puncture. The proof uses the harmonic map heat flow.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:29:41 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Gupta",
"Subhojoy",
""
],
[
"Sau",
"Gobinda",
""
]
] |
2508.10336
|
Etienne Roquain
|
Pierre Humbert and Ulysse Gazin and Ruth Heller and Etienne Roquain
|
Online selective conformal inference: adaptive scores, convergence rate and optimality
| null | null | null | null |
math.ST stat.ML stat.TH
|
http://creativecommons.org/licenses/by/4.0/
|
In a supervised online setting, quantifying uncertainty has been proposed in the seminal work of \cite{gibbs2021adaptive}. For any given point-prediction algorithm, their method (ACI) produces a conformal prediction set with an average missed coverage getting close to a pre-specified level $\alpha$ for a long time horizon. We introduce an extended version of this algorithm, called OnlineSCI, allowing the user to additionally select times where such an inference should be made. OnlineSCI encompasses several prominent online selective tasks, such as building prediction intervals for extreme outcomes, classification with abstention, and online testing. While OnlineSCI controls the average missed coverage on the selected in an adversarial setting, our theoretical results also show that it controls the instantaneous error rate (IER) at the selected times, up to a non-asymptotical remainder term. Importantly, our theory covers the case where OnlineSCI updates the point-prediction algorithm at each time step, a property which we refer to as {\it adaptive} capability. We show that the adaptive versions of OnlineSCI can convergence to an optimal solution and provide an explicit convergence rate in each of the aforementioned application cases, under specific mild conditions. Finally, the favorable behavior of OnlineSCI in practice is illustrated by numerical experiments.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:36:14 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Humbert",
"Pierre",
""
],
[
"Gazin",
"Ulysse",
""
],
[
"Heller",
"Ruth",
""
],
[
"Roquain",
"Etienne",
""
]
] |
2508.10337
|
Chenliang Zhang
|
Chenliang Zhang, Lin Wang, Yuanyuan Lu, Yusheng Qi, Kexin Wang, Peixu Hou, Wenshi Chen
|
A Curriculum Learning Approach to Reinforcement Learning: Leveraging RAG for Multimodal Question Answering
| null | null | null | null |
cs.AI cs.LG
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
This paper describes the solutions of the Dianping-Trust-Safety team for the META CRAG-MM challenge. The challenge requires building a comprehensive retrieval-augmented generation system capable for multi-modal multi-turn question answering. The competition consists of three tasks: (1) answering questions using structured data retrieved from an image-based mock knowledge graph, (2) synthesizing information from both knowledge graphs and web search results, and (3) handling multi-turn conversations that require context understanding and information aggregation from multiple sources. For Task 1, our solution is based on the vision large language model, enhanced by supervised fine-tuning with knowledge distilled from GPT-4.1. We further applied curriculum learning strategies to guide reinforcement learning, resulting in improved answer accuracy and reduced hallucination. For Task 2 and Task 3, we additionally leveraged web search APIs to incorporate external knowledge, enabling the system to better handle complex queries and multi-turn conversations. Our approach achieved 1st place in Task 1 with a significant lead of 52.38\%, and 3rd place in Task 3, demonstrating the effectiveness of the integration of curriculum learning with reinforcement learning in our training pipeline.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:37:56 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Chenliang",
""
],
[
"Wang",
"Lin",
""
],
[
"Lu",
"Yuanyuan",
""
],
[
"Qi",
"Yusheng",
""
],
[
"Wang",
"Kexin",
""
],
[
"Hou",
"Peixu",
""
],
[
"Chen",
"Wenshi",
""
]
] |
2508.10338
|
Bo Wu
|
Bo Wu (1) and Pengfei Zhou (1) ((1) University of Pittsburgh)
|
Near-realtime Earth Observation Via Starlink LEO Satellite Constellation
| null | null | null | null |
cs.NI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Earth observation (EO) satellites in Low Earth Orbit (LEO) are collecting vast amounts of data, which are invaluable for applications such as monitoring forest fires. However, data downloading from EO satellites faces significant challenges due to the limited number of ground stations and the brief communication windows with them. Conversely, emerging LEO constellations like Starlink have enabled continuous connectivity and revolutionized access for ordinary users globally, who can connect via a simple satellite dish. In this paper, we study the feasibility of supporting EO satellites with Starlink satellite infrastructure and introduce a novel data delivery system, designated as "Starlink Space User" (SSU), for relaying data from observation satellites. SSU treats EO satellites as space users of Starlink, facilitating efficient data transfer to Earth. At the core of SSU is a novel class of algorithms designed for link and PoP selection, as well as system scheduling optimization, that operate effectively atop Starlink's proprietary infrastructure. We assess the performance of SSU using trace-driven simulations alongside real-world Starlink performance measurements. Our results demonstrate that the proposed Starlink-aided design can significantly reduce the median backlog (data not delivered) per satellite.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:44:51 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Wu",
"Bo",
"",
"University of Pittsburgh"
],
[
"Zhou",
"Pengfei",
"",
"University of Pittsburgh"
]
] |
2508.10339
|
Andrew Bai
|
Andrew Bai, Justin Cui, Ruochen Wang, Cho-Jui Hsieh
|
Concepts or Skills? Rethinking Instruction Selection for Multi-modal Models
|
11 pages, 1 figure
| null | null | null |
cs.CV cs.LG
|
http://creativecommons.org/licenses/by-sa/4.0/
|
Vision-language instruction tuning achieves two main purposes: learning visual concepts and learning visual skills. In this paper, we found that vision-language benchmarks fall into the dichotomy of mainly benefiting from training on instructions with similar skills or visual concepts. Inspired by the discovery, we designed a simple targeted training data selection method to optimize the performance of a given benchmark. We first extract the concepts/skills from the benchmark, determine whether the benchmark predominantly benefits from similar concepts or skills, and finally select instructions with the most matching concepts/skills. Experiments on 10+ benchmarks validate the effectiveness of our targeted data selection method, showing +0.9\% over the best existing baseline averaged over all benchmarks and +1.5\% on the skill-focused subset. Our findings underscore the importance of recognizing the inherent trade-off within instruction selection, which requires balancing the acquisition of conceptual knowledge against visual skill.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:48:38 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Bai",
"Andrew",
""
],
[
"Cui",
"Justin",
""
],
[
"Wang",
"Ruochen",
""
],
[
"Hsieh",
"Cho-Jui",
""
]
] |
2508.10340
|
Guangyao Shi
|
Chak Lam Shek, Guangyao Shi, Pratap Tokekar
|
Multi-Agent Trust Region Policy Optimisation: A Joint Constraint Approach
| null | null | null | null |
cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Multi-agent reinforcement learning (MARL) requires coordinated and stable policy updates among interacting agents. Heterogeneous-Agent Trust Region Policy Optimization (HATRPO) enforces per-agent trust region constraints using Kullback-Leibler (KL) divergence to stabilize training. However, assigning each agent the same KL threshold can lead to slow and locally optimal updates, especially in heterogeneous settings. To address this limitation, we propose two approaches for allocating the KL divergence threshold across agents: HATRPO-W, a Karush-Kuhn-Tucker-based (KKT-based) method that optimizes threshold assignment under global KL constraints, and HATRPO-G, a greedy algorithm that prioritizes agents based on improvement-to-divergence ratio. By connecting sequential policy optimization with constrained threshold scheduling, our approach enables more flexible and effective learning in heterogeneous-agent settings. Experimental results demonstrate that our methods significantly boost the performance of HATRPO, achieving faster convergence and higher final rewards across diverse MARL benchmarks. Specifically, HATRPO-W and HATRPO-G achieve comparable improvements in final performance, each exceeding 22.5%. Notably, HATRPO-W also demonstrates more stable learning dynamics, as reflected by its lower variance.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:48:46 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Shek",
"Chak Lam",
""
],
[
"Shi",
"Guangyao",
""
],
[
"Tokekar",
"Pratap",
""
]
] |
2508.10341
|
Quanyu Tang
|
Quanyu Tang
|
An interpolation approach to Schoenberg type inequalities
|
9 pages. Comments and suggestions are welcome
| null | null | null |
math.CV math.FA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The classical Schoenberg inequality relates the squared moduli of the critical points of a polynomial to those of its zeros, under the condition that the centroid of the zeros lies at the origin. Its generalizations to other orders are referred to as Schoenberg type inequalities. While higher-order analogues have been established for certain even exponents, the general case remains poorly understood. In this paper, we substantially extend the known results by introducing an interpolation-based framework that unifies the treatment of Schoenberg type inequalities across a continuum of exponents. Our approach employs complex interpolation between $\ell^p$ spaces and Schatten $p$-classes $S_p$, yielding sharp Schoenberg type inequalities of order $p$ for all $p \ge 1$, including previously inaccessible intermediate exponents. This completely resolves an open problem posed by Kushel and Tyaglov. In the process, we also provide a new proof of the Schoenberg type inequality of order $1$.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:51:45 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Tang",
"Quanyu",
""
]
] |
2508.10342
|
Bang Quan Zheng
|
Bang Quan Zheng
|
Identifying Unmeasured Confounders in Panel Causal Models: A Two-Stage LM-Wald Approach
| null | null | null | null |
stat.ME
|
http://creativecommons.org/licenses/by/4.0/
|
Panel data are widely used in political science to draw causal inferences. However, these models often rely on the strong and untested assumption of sequential ignorability--that no unmeasured variables influence both the independent and outcome variables across time. Grounded in psychometric literature on latent variable modeling, this paper introduces the Two-Stage LM-Wald (2SLW) approach, a diagnostic tool that extends the Lagrange Multiplier (LM) and Wald tests to detect violations of this assumption in panel causal models. Using Monte Carlo simulations within the Random Intercept Cross-Lagged Panel Model (RI-CLPM), which separates within and between person effects, I demonstrate the 2SLW's ability to detect unmeasured confounding across three key scenarios: biased corrections, distorted direct effects, and altered mediation pathways. I also illustrate the approach with an empirical application to real-world panel data. By providing a practical and theoretically grounded diagnostic, the 2SLW approach enhances the robustness of causal inferences in the presence of potential time-varying confounders. Moreover, it can be readily implemented using the R package lavaan.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:55:56 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zheng",
"Bang Quan",
""
]
] |
2508.10343
|
Boris Gutman
|
Daniel Semchin, Emile d'Angremont, Marco Lorenzi, and Boris Gutman
|
Scalable Modeling of Nonlinear Network Dynamics in Neurodegenerative Disease
| null | null | null | null |
q-bio.QM q-bio.NC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Mechanistic models of progressive neurodegeneration offer great potential utility for clinical use and novel treatment development. Toward this end, several connectome-informed models of neuroimaging biomarkers have been proposed. However, these models typically do not scale well beyond a small number of biomarkers due to heterogeneity in individual disease trajectories and a large number of parameters. To address this, we introduce the Connectome-based Monotonic Inference of Neurodegenerative Dynamics (COMIND). The model combines concepts from diffusion and logistic models with structural brain connectivity. This guarantees monotonic disease trajectories while maintaining a limited number of parameters to improve scalability. We evaluate our model on simulated data as well as on the Parkinson's Progressive Markers Initiative (PPMI) data. Our model generalizes to anatomical imaging representations from a standard brain atlas without the need to reduce biomarker number.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 04:58:26 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Semchin",
"Daniel",
""
],
[
"d'Angremont",
"Emile",
""
],
[
"Lorenzi",
"Marco",
""
],
[
"Gutman",
"Boris",
""
]
] |
2508.10344
|
Thomas Hangelbroek
|
Thomas Hangelbroek and Christian Rieger and Grady B. Wright
|
A Semi-Lagrangian scheme on embedded manifolds using generalized local polynomial reproductions
| null | null | null | null |
math.NA cs.NA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We analyze rates of uniform convergence for a class of high-order semi-Lagrangian schemes for first-order, time-dependent partial differential equations on embedded submanifolds of $\mathbb{R}^d$ (including advection equations on surfaces) by extending the error analysis of Falcone and Ferretti. A central requirement in our analysis is a remapping operator that achieves both high approximation orders and strong stability, a combination that is challenging to obtain and of independent interest. For this task, we propose a novel mesh-free remapping operator based on $\ell_1$ minimizing generalized polynomial reproduction, which uses only point values and requires no additional geometric information from the manifold (such as access to tangent spaces or curvature). Our framework also rigorously addresses the numerical solution of ordinary differential equations on manifolds via projection methods. We include numerical experiments that support the theoretical results and also suggest some new directions for future research.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:02:18 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Hangelbroek",
"Thomas",
""
],
[
"Rieger",
"Christian",
""
],
[
"Wright",
"Grady B.",
""
]
] |
2508.10345
|
Claire Jie Zhang
|
Claire Jie Zhang, Seyed A. Esmaeili, Jamie Morgenstern
|
Welfare-Centric Clustering
| null | null | null | null |
cs.LG cs.AI cs.CY cs.DS
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Fair clustering has traditionally focused on ensuring equitable group representation or equalizing group-specific clustering costs. However, Dickerson et al. (2025) recently showed that these fairness notions may yield undesirable or unintuitive clustering outcomes and advocated for a welfare-centric clustering approach that models the utilities of the groups. In this work, we model group utilities based on both distances and proportional representation and formalize two optimization objectives based on welfare-centric clustering: the Rawlsian (Egalitarian) objective and the Utilitarian objective. We introduce novel algorithms for both objectives and prove theoretical guarantees for them. Empirical evaluations on multiple real-world datasets demonstrate that our methods significantly outperform existing fair clustering baselines.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:02:32 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Claire Jie",
""
],
[
"Esmaeili",
"Seyed A.",
""
],
[
"Morgenstern",
"Jamie",
""
]
] |
2508.10346
|
Md Ashraf Uddin
|
Md Ashraf Uddin, Nam H. Chu, Reza Rafeh
|
A Hierarchical IDS for Zero-Day Attack Detection in Internet of Medical Things Networks
|
13 pages, and 4 figures
| null | null | null |
cs.LG cs.NI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The Internet of Medical Things (IoMT) is driving a healthcare revolution but remains vulnerable to cyberattacks such as denial of service, ransomware, data hijacking, and spoofing. These networks comprise resource constrained, heterogeneous devices (e.g., wearable sensors, smart pills, implantables), making traditional centralized Intrusion Detection Systems (IDSs) unsuitable due to response delays, privacy risks, and added vulnerabilities. Centralized IDSs require all sensors to transmit data to a central server, causing delays or network disruptions in dense environments. Running IDSs locally on IoMT devices is often infeasible due to limited computation, and even lightweight IDS components remain at risk if updated models are delayed leaving them exposed to zero-day attacks that threaten patient health and data security. We propose a multi level IoMT IDS framework capable of detecting zero day attacks and distinguishing between known and unknown threats. The first layer (near Edge) filters traffic at a coarse level (attack or not) using meta-learning or One Class Classification (OCC) with the usfAD algorithm. Subsequent layers (far Edge, Cloud) identify attack type and novelty. Experiments on the CICIoMT2024 dataset show 99.77 percentage accuracy and 97.8 percentage F1-score. The first layer detects zero-day attacks with high accuracy without needing new datasets, ensuring strong applicability in IoMT environments. Additionally, the meta-learning approach achieves high.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:08:37 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Uddin",
"Md Ashraf",
""
],
[
"Chu",
"Nam H.",
""
],
[
"Rafeh",
"Reza",
""
]
] |
2508.10347
|
Charis Tsikkou
|
Josh Culver, Aubrey Ayres, Evan Halloran, Ryan Lin, Emily Peng, Charis Tsikkou
|
An Analysis of the Riemann Problem for a $2 \times 2$ System of Keyfitz-Kranzer Type Balance Laws With a Time-Dependent Source Term
|
35 pages. arXiv admin note: text overlap with arXiv:2508.05927
| null | null | null |
math.AP math-ph math.CA math.DS math.MP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We consider a system consisting of one conservation law and one balance law with a time-dependent source term, and provide a comprehensive analysis of Riemann solutions, including the non-classical overcompressive delta shocks. The minimal yet representative structure of the system captures essential features of transport under density constraints and, despite its simplicity, serves as a versatile prototype for crowd-limited transport processes across diverse contexts, including biological aggregation, ecological dispersal, granular compaction, and traffic congestion. In addition to non-self-similar solutions mentioned above, the associated Riemann problem admits solution structures that traverse vacuum states ($\rho = 0$) and the critical density threshold ($\rho = \bar{\rho}$), where mobility vanishes and characteristic speed degenerates. Moreover, the explicit time dependence in the source term leads to the breakdown of self-similarity, resulting in distinct Riemann solutions over successive time intervals and highlighting the dynamic nature of the solution landscape. The theoretical findings are numerically confirmed using the Local Lax-Friedrichs scheme.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:10:04 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Culver",
"Josh",
""
],
[
"Ayres",
"Aubrey",
""
],
[
"Halloran",
"Evan",
""
],
[
"Lin",
"Ryan",
""
],
[
"Peng",
"Emily",
""
],
[
"Tsikkou",
"Charis",
""
]
] |
2508.10348
|
Tung T. Nguyen
|
Tung T. Nguyen, Nguyen Duy T\^an
|
Supercharacters of finite abelian groups and applications to spectra of $U$-unitary Cayley graphs
|
Comments are welcome!
| null | null | null |
math.NT
|
http://creativecommons.org/licenses/by/4.0/
|
We define super-Cayley graphs over a finite abelian group $G$. Using the theory of supercharacters on $G$, we explain how their spectra can be realized as a super-Fourier transform of a superclass characteristic function. Consequently, we show that a super-Cayley graph is determined by its spectrum once an indexing on the underlying group $G$ is fixed. This generalizes a theorem by Sander-Sander, which investigates the case where $G$ is a cyclic group. We then use our theory to define and study the concept of a $U$-unitary Cayley graph over a finite commutative ring $R$, where $U$ is a subgroup of the unit group of $R$. Furthermore, when the underlying ring is a Frobenius ring, we show that there is a natural supercharacter theory associated with $U$. By applying the general theory of super-Cayley graphs developed in the first part, we explore various spectral properties of these $U$-unitary Cayley graphs, including their rationality and connections to various arithmetical sums.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:11:30 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Nguyen",
"Tung T.",
""
],
[
"Tân",
"Nguyen Duy",
""
]
] |
2508.10349
|
Jiaxiang Geng
|
Tianjun Yuan, Jiaxiang Geng, Pengchao Han, Xianhao Chen, Bing Luo
|
Flexible Personalized Split Federated Learning for On-Device Fine-Tuning of Foundation Models
|
10 pages, Submitted to INFOCOM2026
| null | null | null |
cs.DC cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Fine-tuning foundation models is critical for superior performance on personalized downstream tasks, compared to using pre-trained models. Collaborative learning can leverage local clients' datasets for fine-tuning, but limited client data and heterogeneous data distributions hinder effective collaboration. To address the challenge, we propose a flexible personalized federated learning paradigm that enables clients to engage in collaborative learning while maintaining personalized objectives. Given the limited and heterogeneous computational resources available on clients, we introduce \textbf{flexible personalized split federated learning (FlexP-SFL)}. Based on split learning, FlexP-SFL allows each client to train a portion of the model locally while offloading the rest to a server, according to resource constraints. Additionally, we propose an alignment strategy to improve personalized model performance on global data. Experimental results show that FlexP-SFL outperforms baseline models in personalized fine-tuning efficiency and final accuracy.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:14:00 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Yuan",
"Tianjun",
""
],
[
"Geng",
"Jiaxiang",
""
],
[
"Han",
"Pengchao",
""
],
[
"Chen",
"Xianhao",
""
],
[
"Luo",
"Bing",
""
]
] |
2508.10350
|
Samer Lahoud
|
Samer Lahoud and Kinda Khawam
|
Semantic Communication with Distribution Learning through Sequential Observations
| null | null | null | null |
cs.LG cs.NI
|
http://creativecommons.org/licenses/by/4.0/
|
Semantic communication aims to convey meaning rather than bit-perfect reproduction, representing a paradigm shift from traditional communication. This paper investigates distribution learning in semantic communication where receivers must infer the underlying meaning distribution through sequential observations. While semantic communication traditionally optimizes individual meaning transmission, we establish fundamental conditions for learning source statistics when priors are unknown. We prove that learnability requires full rank of the effective transmission matrix, characterize the convergence rate of distribution estimation, and quantify how estimation errors translate to semantic distortion. Our analysis reveals a fundamental trade-off: encoding schemes optimized for immediate semantic performance often sacrifice long-term learnability. Experiments on CIFAR-10 validate our theoretical framework, demonstrating that system conditioning critically impacts both learning rate and achievable performance. These results provide the first rigorous characterization of statistical learning in semantic communication and offer design principles for systems that balance immediate performance with adaptation capability.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:15:05 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Lahoud",
"Samer",
""
],
[
"Khawam",
"Kinda",
""
]
] |
2508.10351
|
Lei Cao
|
Zhentai Zhang, Danyi Weng, Guibin Zhang, Xiang Chen, Kaixing Long, Jian Geng, Yanmeng Lu, Lei Zhang, Zhitao Zhou, Lei Cao
|
Glo-DMU: A Deep Morphometry Framework of Ultrastructural Characterization in Glomerular Electron Microscopic Images
|
15 pages, 6 figures
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Complex and diverse ultrastructural features can indicate the type, progression, and prognosis of kidney diseases. Recently, computational pathology combined with deep learning methods has shown tremendous potential in advancing automatic morphological analysis of glomerular ultrastructure. However, current research predominantly focuses on the recognition of individual ultrastructure, which makes it challenging to meet practical diagnostic needs. In this study, we propose the glomerular morphometry framework of ultrastructural characterization (Glo-DMU), which is grounded on three deep models: the ultrastructure segmentation model, the glomerular filtration barrier region classification model, and the electron-dense deposits detection model. Following the conventional protocol of renal biopsy diagnosis, this framework simultaneously quantifies the three most widely used ultrastructural features: the thickness of glomerular basement membrane, the degree of foot process effacement, and the location of electron-dense deposits. We evaluated the 115 patients with 9 renal pathological types in real-world diagnostic scenarios, demonstrating good consistency between automatic quantification results and morphological descriptions in the pathological reports. Glo-DMU possesses the characteristics of full automation, high precision, and high throughput, quantifying multiple ultrastructural features simultaneously, and providing an efficient tool for assisting renal pathologists.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:25:30 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Zhentai",
""
],
[
"Weng",
"Danyi",
""
],
[
"Zhang",
"Guibin",
""
],
[
"Chen",
"Xiang",
""
],
[
"Long",
"Kaixing",
""
],
[
"Geng",
"Jian",
""
],
[
"Lu",
"Yanmeng",
""
],
[
"Zhang",
"Lei",
""
],
[
"Zhou",
"Zhitao",
""
],
[
"Cao",
"Lei",
""
]
] |
2508.10352
|
Beso Mikaberidze
|
Beso Mikaberidze, Teimuraz Saghinadze, Simon Ostermann, Philipp Muller
|
Cross-Prompt Encoder for Low-Performing Languages
| null | null | null | null |
cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
Soft prompts have emerged as a powerful alternative to adapters in parameter-efficient fine-tuning (PEFT), enabling large language models (LLMs) to adapt to downstream tasks without architectural changes or parameter updates. While prior work has focused on stabilizing training via parameter interaction in small neural prompt encoders, their broader potential for transfer across languages remains unexplored. In this paper, we demonstrate that a prompt encoder can play a central role in improving performance on low-performing languages-those that achieve poor accuracy even under full-model fine-tuning. We introduce the Cross-Prompt Encoder (XPE), which combines a lightweight encoding architecture with multi-source training on typologically diverse languages - a design that enables the model to capture abstract and transferable patterns across languages. To complement XPE, we propose a Dual Soft Prompt mechanism that combines an encoder-based prompt with a directly trained standard soft prompt. This hybrid design proves especially effective for target languages that benefit from both broadly shared structure and language-specific alignment. Experiments on the SIB-200 benchmark reveal a consistent trade-off: XPE is most effective for low-performing languages, while hybrid variants offer broader adaptability across multilingual settings.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:36:21 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Mikaberidze",
"Beso",
""
],
[
"Saghinadze",
"Teimuraz",
""
],
[
"Ostermann",
"Simon",
""
],
[
"Muller",
"Philipp",
""
]
] |
2508.10353
|
Kalyan Ramana Gattoz Mr.
|
G. Kalyan Ramana, Sumit Yempalle and Prasad S. Onkar
|
Mental Effort Estimation in Motion Exploration and Concept Generation Design Tasks using Inter-Band Relative Power Difference of EEG
| null | null | null | null |
cs.HC stat.ME
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Conceptual design is a cognitively complex task, especially in the engineering design of products having relative motion between components. Designers prefer sketching as a medium for conceptual design and use gestures and annotations to represent such relative motion. Literature suggests that static representations of motion in sketches may not achieve the intended functionality when realised, because it primarily depends on the designers' mental capabilities for motion simulation. Thus, it is important to understand the cognitive phenomena when designers are exploring concepts of articulated products. The current work is an attempt to understand design neurocognition by categorising the tasks and measuring the mental effort involved in these tasks using EEG. The analysis is intended to validate design intervention tools to support the conceptual design involving motion exploration. A novel EEG-based metric, inter-Band Relative Power Difference (inter-BRPD), is introduced to quantify mental effort. A design experiment is conducted with 32 participants, where they have to perform one control task and 2 focus tasks corresponding to the motion exploration task (MET) and the concept generation task (CGT), respectively. EEG data is recorded during the 3 tasks, cleaned, processed and analysed using the MNE library in Python. It is observed from the results that inter-BRPD captures the essence of mental effort with half the number of conventionally used parameters. The reliability and efficacy of the inter-BRPD metric are also statistically validated against literature-based cognitive metrics. With these new insights, the study opens up possibilities for creating support for conceptual design and its evaluation.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:43:54 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Ramana",
"G. Kalyan",
""
],
[
"Yempalle",
"Sumit",
""
],
[
"Onkar",
"Prasad S.",
""
]
] |
2508.10354
|
Antonino Ficarra
|
Antonino Ficarra, Somayeh Moradi
|
Stanley-Reisner ideals with linear powers
| null | null | null | null |
math.AC math.CO
|
http://creativecommons.org/licenses/by-sa/4.0/
|
Let $S = K[x_1, \dots, x_n]$ be the standard graded polynomial ring over a field $K$. In this paper, we address and completely solve two fundamental open questions in Commutative Algebra:
(i) For which degrees $d$, does there exist a uniform combinatorial characterization of all squarefree monomial ideals in $S$ having $d$-linear resolutions?
(ii) For which degrees $d$, does having a linear resolution coincide with having linear powers for all squarefree monomial ideals of $S$ generated in degree $d$?
Let $\mathcal{I}_{n,d}(K)$ denote the class of squarefree monomial ideals of $S$ having a $d$-linear resolution. Our main result establishes the equivalence of the following conditions:
(a) Any squarefree monomial ideal $I$ in $S$ generated in degree $d$ has a linear resolution, if and only if, $I$ has linear powers.
(b) $\mathcal{I}_{n,d}(K)$ is independent of the base field $K$.
(c) $d\in\{0,1,2,n{-}2,n{-}1,n\}$.
In each of these degrees, we show that a squarefree monomial ideal has a linear resolution if and only if all of its powers admit linear quotients, and we combinatorially classify such ideals. In contrast, for each degree $3\le d\le n{-}3$, we construct fully-supported squarefree monomial ideals $I$ and $J$ in $S$ generated in degree $d$ such that the linear resolution property of $I$ depends on the choice of the base field, $J$ has a linear resolution and $J^2$ does not have a linear resolution.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:47:48 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Ficarra",
"Antonino",
""
],
[
"Moradi",
"Somayeh",
""
]
] |
2508.10355
|
Sang Park
|
Jungyup Lee, Jemin Kim, Sang Park, SeungJae Lee
|
Making Qwen3 Think in Korean with Reinforcement Learning
| null | null | null | null |
cs.CL
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
We present a two-stage fine-tuning approach to make the large language model Qwen3 14B "think" natively in Korean. In the first stage, supervised fine-tuning (SFT) on a high-quality Korean reasoning dataset establishes a strong foundation in Korean logical reasoning, yielding notable improvements in Korean-language tasks and even some gains in general reasoning ability. In the second stage, we employ reinforcement learning with a customized Group Relative Policy Optimization (GRPO) algorithm to further enhance both Korean reasoning alignment and overall problem-solving performance. We address critical stability challenges in GRPO training - such as reward hacking and policy collapse - by introducing an oracle judge model that calibrates the reward signal. Our approach achieves stable learning (avoiding the collapse observed in naive GRPO) and leads to steady, incremental performance gains. The final RL-tuned model demonstrates substantially improved results on advanced reasoning benchmarks (particularly math and coding tasks) while maintaining knowledge and language proficiency, successfully conducting its internal chain-of-thought entirely in Korean.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:49:34 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Lee",
"Jungyup",
""
],
[
"Kim",
"Jemin",
""
],
[
"Park",
"Sang",
""
],
[
"Lee",
"SeungJae",
""
]
] |
2508.10356
|
Khondoker Ittehadul Islam
|
Hylke Westerdijk, Ben Blankenborg, Khondoker Ittehadul Islam
|
Improving OCR for Historical Texts of Multiple Languages
| null | null | null | null |
cs.CV cs.CL
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
This paper presents our methodology and findings from three tasks across Optical Character Recognition (OCR) and Document Layout Analysis using advanced deep learning techniques. First, for the historical Hebrew fragments of the Dead Sea Scrolls, we enhanced our dataset through extensive data augmentation and employed the Kraken and TrOCR models to improve character recognition. In our analysis of 16th to 18th-century meeting resolutions task, we utilized a Convolutional Recurrent Neural Network (CRNN) that integrated DeepLabV3+ for semantic segmentation with a Bidirectional LSTM, incorporating confidence-based pseudolabeling to refine our model. Finally, for modern English handwriting recognition task, we applied a CRNN with a ResNet34 encoder, trained using the Connectionist Temporal Classification (CTC) loss function to effectively capture sequential dependencies. This report offers valuable insights and suggests potential directions for future research.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:52:14 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Westerdijk",
"Hylke",
""
],
[
"Blankenborg",
"Ben",
""
],
[
"Islam",
"Khondoker Ittehadul",
""
]
] |
2508.10357
|
Xiudi Li
|
Xiudi Li and Sijia Li
|
Efficient Inference for Time-to-Event Outcomes by Integrating Right-Censored and Current Status Data
| null | null | null | null |
stat.ME
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We propose a semiparametric data fusion framework for efficient inference on survival probabilities by integrating right-censored and current status data. Existing data fusion methods focus largely on fusing right-censored data only, while standard meta-analysis approaches are inadequate for combining right-censored and current status data, as estimators based on current status data alone typically converge at slower rates and have non-normal limiting distributions. In this work, we consider a semiparametric model under exchangeable event time distribution across data sources. We derive the canonical gradient of the survival probability at a given time, and develop a one-step estimator along with the corresponding inference procedure. Our proposed estimator is doubly robust and attains the semiparametric efficiency bound under mild conditions. Importantly, we show that incorporating current status data can lead to meaningful efficiency gains despite the slower convergence rate of current status-only estimators. We demonstrate the performance of our proposed method in simulations and discuss extensions to settings with covariate shift. We believe that this work has the potential to open new directions in data fusion methodology, particularly for settings involving mixed censoring types.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:55:07 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Li",
"Xiudi",
""
],
[
"Li",
"Sijia",
""
]
] |
2508.10358
|
Mengtao Zhou
|
Mengtao Zhou, Sifan Wu, Huan Zhang, Qi Sima, Bang Liu
|
What to Ask Next? Probing the Imaginative Reasoning of LLMs with TurtleSoup Puzzles
| null | null | null | null |
cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We investigate the capacity of Large Language Models (LLMs) for imaginative reasoning--the proactive construction, testing, and revision of hypotheses in information-sparse environments. Existing benchmarks, often static or focused on social deduction, fail to capture the dynamic, exploratory nature of this reasoning process. To address this gap, we introduce a comprehensive research framework based on the classic "Turtle Soup" game, integrating a benchmark, an agent, and an evaluation protocol. We present TurtleSoup-Bench, the first large-scale, bilingual, interactive benchmark for imaginative reasoning, comprising 800 turtle soup puzzles sourced from both the Internet and expert authors. We also propose Mosaic-Agent, a novel agent designed to assess LLMs' performance in this setting. To evaluate reasoning quality, we develop a multi-dimensional protocol measuring logical consistency, detail completion, and conclusion alignment. Experiments with leading LLMs reveal clear capability limits, common failure patterns, and a significant performance gap compared to humans. Our work offers new insights into LLMs' imaginative reasoning and establishes a foundation for future research on exploratory agent behavior.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:55:42 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhou",
"Mengtao",
""
],
[
"Wu",
"Sifan",
""
],
[
"Zhang",
"Huan",
""
],
[
"Sima",
"Qi",
""
],
[
"Liu",
"Bang",
""
]
] |
2508.10359
|
Hao Wang
|
Hao Wang, Hongkui Zheng, Kai He, Abolfazl Razi
|
AtomDiffuser: Time-Aware Degradation Modeling for Drift and Beam Damage in STEM Imaging
| null | null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Scanning transmission electron microscopy (STEM) plays a critical role in modern materials science, enabling direct imaging of atomic structures and their evolution under external interferences. However, interpreting time-resolved STEM data remains challenging due to two entangled degradation effects: spatial drift caused by mechanical and thermal instabilities, and beam-induced signal loss resulting from radiation damage. These factors distort both geometry and intensity in complex, temporally correlated ways, making it difficult for existing methods to explicitly separate their effects or model material dynamics at atomic resolution. In this work, we present AtomDiffuser, a time-aware degradation modeling framework that disentangles sample drift and radiometric attenuation by predicting an affine transformation and a spatially varying decay map between any two STEM frames. Unlike traditional denoising or registration pipelines, our method leverages degradation as a physically heuristic, temporally conditioned process, enabling interpretable structural evolutions across time. Trained on synthetic degradation processes, AtomDiffuser also generalizes well to real-world cryo-STEM data. It further supports high-resolution degradation inference and drift alignment, offering tools for visualizing and quantifying degradation patterns that correlate with radiation-induced atomic instabilities.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:56:31 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Wang",
"Hao",
""
],
[
"Zheng",
"Hongkui",
""
],
[
"He",
"Kai",
""
],
[
"Razi",
"Abolfazl",
""
]
] |
2508.10360
|
Henry Zhong
|
Henry Zhong, J\"org M. Buchholz, Julian Maclaren, Simon Carlile, and Richard Lyon
|
A dataset and model for recognition of audiologically relevant environments for hearing aids: AHEAD-DS and YAMNet+
| null | null | null | null |
cs.SD eess.AS
|
http://creativecommons.org/licenses/by/4.0/
|
Scene recognition of audiologically relevant environments is important for hearing aids; however, it is challenging, in part because of the limitations of existing datasets. Datasets often lack public accessibility, completeness, or audiologically relevant labels, hindering systematic comparison of machine learning models. Deploying these models on resource-constrained edge devices presents another challenge. Our solution is two-fold: we leverage several open source datasets to create AHEAD-DS, a dataset designed for scene recognition of audiologically relevant environments, and introduce YAMNet+, a sound recognition model. AHEAD-DS aims to provide a standardised, publicly available dataset with consistent labels relevant to hearing aids, facilitating model comparison. YAMNet+ is designed for deployment on edge devices like smartphones connected to hearing devices, such as hearing aids and wireless earphones with hearing aid functionality; serving as a baseline model for sound-based scene recognition. YAMNet+ achieved a mean average precision of 0.83 and accuracy of 0.93 on the testing set of AHEAD-DS across fourteen categories of audiologically relevant environments. We found that applying transfer learning from the pretrained YAMNet model was essential. We demonstrated real-time sound-based scene recognition capabilities on edge devices by deploying YAMNet+ to an Android smartphone. Even with a Google Pixel 3 (a phone with modest specifications, released in 2018), the model processes audio with approximately 50ms of latency to load the model, and an approximate linear increase of 30ms per 1 second of audio. Our website and code https://github.com/Australian-Future-Hearing-Initiative .
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:59:21 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhong",
"Henry",
""
],
[
"Buchholz",
"Jörg M.",
""
],
[
"Maclaren",
"Julian",
""
],
[
"Carlile",
"Simon",
""
],
[
"Lyon",
"Richard",
""
]
] |
2508.10361
|
Kohei Kobayashi
|
Kohei Kobayashi
|
Saturable Quantum Speed Limits for Imaginary-Time Evolution
|
7 pages, 0 figure
| null | null | null |
quant-ph
|
http://creativecommons.org/licenses/by-sa/4.0/
|
We derive a Geometric quantum speed limit (QSL) for imaginary-time evolution, where the dynamics is governed by a non-unitary Schr\"{o}dinger equation. By introducing a cost function based on the angular distance between the normalized evolving state and the initial state, we obtain a lower bound on the evolution time expressed as the ratio between this angle and the time-averaged energy dispersion. Our bound is analytical, general, and applicable to arbitrary time-independent Hamiltonians. We analytically evaluate this bound for two physically motivated cases. First, we apply it to a two-level system and derive an expression for the minimal time. Second, we analyze the imaginary-time version of Grover search problem and rigorously reproduce the well-known logarithmic scaling $T=\mathcal{ O}(\log N)$ within our QSL framework.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 05:59:30 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Kobayashi",
"Kohei",
""
]
] |
2508.10362
|
Michael Sun Dr
|
Alex Qiu, Tanish Sarathy, Spencer Nicklin, Michael Sun
|
Understanding Fermat's Last Theorem's Proofs
|
126 pages, feedback and submission recommendations welcome, highlighted text removed
| null | null | null |
math.GM
|
http://creativecommons.org/licenses/by/4.0/
|
We take the perspective of an advanced high school student trying to understand the proof of Fermat's Last Theorem for the first time. We collect definitions and statements needed to summarise how Fermat's Last Theorem was first proved in 1995 as well as to see how the argument has simplified since then. We include a current timeline, outlines of proofs, background material and recent developments, as well as to organise the content in a way that is beginner friendly, offering a preview of what students may expect to learn more deeply in the future.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:00:31 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Qiu",
"Alex",
""
],
[
"Sarathy",
"Tanish",
""
],
[
"Nicklin",
"Spencer",
""
],
[
"Sun",
"Michael",
""
]
] |
2508.10363
|
Donipolo Ghimire
|
Donipolo Ghimire, Aamodh Suresh, Carlos Nieto-Granda, Solmaz S. Kia
|
BEASST: Behavioral Entropic Gradient based Adaptive Source Seeking for Mobile Robots
| null | null | null | null |
cs.RO
|
http://creativecommons.org/licenses/by/4.0/
|
This paper presents BEASST (Behavioral Entropic Gradient-based Adaptive Source Seeking for Mobile Robots), a novel framework for robotic source seeking in complex, unknown environments. Our approach enables mobile robots to efficiently balance exploration and exploitation by modeling normalized signal strength as a surrogate probability of source location. Building on Behavioral Entropy(BE) with Prelec's probability weighting function, we define an objective function that adapts robot behavior from risk-averse to risk-seeking based on signal reliability and mission urgency. The framework provides theoretical convergence guarantees under unimodal signal assumptions and practical stability under bounded disturbances. Experimental validation across DARPA SubT and multi-room scenarios demonstrates that BEASST consistently outperforms state-of-the-art methods, achieving 15% reduction in path length and 20% faster source localization through intelligent uncertainty-driven navigation that dynamically transitions between aggressive pursuit and cautious exploration.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:01:20 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Ghimire",
"Donipolo",
""
],
[
"Suresh",
"Aamodh",
""
],
[
"Nieto-Granda",
"Carlos",
""
],
[
"Kia",
"Solmaz S.",
""
]
] |
2508.10364
|
Xueer Lin
|
Xueer Lin, Chenyu Li, Yuhan Lyu, Zhicong Lu, Zhenhui Peng
|
"Here Comes the Makeup Tutorial You Asked For!": Exploring Communication Strategies and Viewer Engagement in Beauty Videos on Rednote
| null | null | null | null |
cs.HC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
More and more people, especially females, create and view beauty videos covering topics like makeup tutorials and vlogs on social media platforms. Understanding the communication strategies that creators use in these videos and how they affect viewers' engagement can help spread beauty knowledge. By coding 352 beauty videos in Rednote, this study presents a comprehensive taxonomy of communication strategies used by the creators, such as using home as the video background and displaying makeup effects when starting the narrative at the beginning. We further label and computationally classify six categories of comments that reveal viewers' engagement with beauty videos. The regression analyses reveal the effects of beauty video communication strategies on viewers' engagement; for example, calling viewers to take action at the end tends to attract more comments that debate the product's efficacy. We discuss insights into fostering the creation of beauty videos and the communication of beauty knowledge.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:01:42 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Lin",
"Xueer",
""
],
[
"Li",
"Chenyu",
""
],
[
"Lyu",
"Yuhan",
""
],
[
"Lu",
"Zhicong",
""
],
[
"Peng",
"Zhenhui",
""
]
] |
2508.10365
|
Sachin S Sharma
|
Suresh Govindarajan, Sachin S. Sharma, Sankaran Viswanath
|
The affine Brylinski filtration and $\mathscr{W}$-algebras
|
Comments and suggestions are welcome
| null | null | null |
math.RT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The Brylinski-Kostant filtration on a representation of a finite-dimensional semisimple Lie algebra has interpretations in terms of the algebra, geometry and combinatorics of the representation. Its extension to affine Lie algebras was first studied by Slofstra. Recent work of the present authors constructed a Poincar\'{e}-Birkhoff-Witt type basis for the dominant weight spaces of the basic representation of affine Lie algebras of type $A$, which is compatible with the affine Brylinski filtration. In this paper, we overcome the constraint of type dependence, and furnish a new, uniform proof which holds for all simply-laced affine Lie algebras.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:05:28 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Govindarajan",
"Suresh",
""
],
[
"Sharma",
"Sachin S.",
""
],
[
"Viswanath",
"Sankaran",
""
]
] |
2508.10366
|
Jakub \v{S}m\'id
|
Jakub \v{S}m\'id, Pavel P\v{r}ib\'a\v{n}, Pavel Kr\'al
|
Advancing Cross-lingual Aspect-Based Sentiment Analysis with LLMs and Constrained Decoding for Sequence-to-Sequence Models
|
Published in Proceedings of the 17th International Conference on Agents and Artificial Intelligence - Volume 2 (ICAART 2025). Official version: https://www.scitepress.org/Link.aspx?doi=10.5220/0013349400003890
| null |
10.5220/0013349400003890
| null |
cs.CL
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Aspect-based sentiment analysis (ABSA) has made significant strides, yet challenges remain for low-resource languages due to the predominant focus on English. Current cross-lingual ABSA studies often centre on simpler tasks and rely heavily on external translation tools. In this paper, we present a novel sequence-to-sequence method for compound ABSA tasks that eliminates the need for such tools. Our approach, which uses constrained decoding, improves cross-lingual ABSA performance by up to 10\%. This method broadens the scope of cross-lingual ABSA, enabling it to handle more complex tasks and providing a practical, efficient alternative to translation-dependent techniques. Furthermore, we compare our approach with large language models (LLMs) and show that while fine-tuned multilingual LLMs can achieve comparable results, English-centric LLMs struggle with these tasks.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:07:43 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Šmíd",
"Jakub",
""
],
[
"Přibáň",
"Pavel",
""
],
[
"Král",
"Pavel",
""
]
] |
2508.10367
|
Pablo Hern\'adez-C\'amara
|
Pablo Hern\'andez-C\'amara, Alexandra Gomez-Villa, Jose Manuel Ja\'en-Lorites, Jorge Vila-Tom\'as, Jesus Malo, Valero Laparra
|
Contrast Sensitivity Function of Multimodal Vision-Language Models
| null | null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Assessing the alignment of multimodal vision-language models~(VLMs) with human perception is essential to understand how they perceive low-level visual features. A key characteristic of human vision is the contrast sensitivity function (CSF), which describes sensitivity to spatial frequency at low-contrasts. Here, we introduce a novel behavioral psychophysics-inspired method to estimate the CSF of chat-based VLMs by directly prompting them to judge pattern visibility at different contrasts for each frequency. This methodology is closer to the real experiments in psychophysics than the previously reported. Using band-pass filtered noise images and a diverse set of prompts, we assess model responses across multiple architectures. We find that while some models approximate human-like CSF shape or magnitude, none fully replicate both. Notably, prompt phrasing has a large effect on the responses, raising concerns about prompt stability. Our results provide a new framework for probing visual sensitivity in multimodal models and reveal key gaps between their visual representations and human perception.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:07:44 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Hernández-Cámara",
"Pablo",
""
],
[
"Gomez-Villa",
"Alexandra",
""
],
[
"Jaén-Lorites",
"Jose Manuel",
""
],
[
"Vila-Tomás",
"Jorge",
""
],
[
"Malo",
"Jesus",
""
],
[
"Laparra",
"Valero",
""
]
] |
2508.10368
|
Jakub \v{S}m\'id
|
V\'aclav Tran, Jakub \v{S}m\'id, Ji\v{r}\'i Mart\'inek, Ladislav Lenc, Pavel Kr\'al
|
Large Language Models for Summarizing Czech Historical Documents and Beyond
|
Published in Proceedings of the 17th International Conference on Agents and Artificial Intelligence - Volume 2 (ICAART 2025). Official version: https://www.scitepress.org/Link.aspx?doi=10.5220/0013374100003890
| null |
10.5220/0013374100003890
| null |
cs.CL
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Text summarization is the task of shortening a larger body of text into a concise version while retaining its essential meaning and key information. While summarization has been significantly explored in English and other high-resource languages, Czech text summarization, particularly for historical documents, remains underexplored due to linguistic complexities and a scarcity of annotated datasets. Large language models such as Mistral and mT5 have demonstrated excellent results on many natural language processing tasks and languages. Therefore, we employ these models for Czech summarization, resulting in two key contributions: (1) achieving new state-of-the-art results on the modern Czech summarization dataset SumeCzech using these advanced models, and (2) introducing a novel dataset called Posel od \v{C}erchova for summarization of historical Czech documents with baseline results. Together, these contributions provide a great potential for advancing Czech text summarization and open new avenues for research in Czech historical text processing.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:07:49 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Tran",
"Václav",
""
],
[
"Šmíd",
"Jakub",
""
],
[
"Martínek",
"Jiří",
""
],
[
"Lenc",
"Ladislav",
""
],
[
"Král",
"Pavel",
""
]
] |
2508.10369
|
Jakub \v{S}m\'id
|
Jakub \v{S}m\'id, Pavel P\v{r}ib\'a\v{n}, Pavel Kr\'al
|
Improving Generative Cross-lingual Aspect-Based Sentiment Analysis with Constrained Decoding
| null | null | null | null |
cs.CL
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
While aspect-based sentiment analysis (ABSA) has made substantial progress, challenges remain for low-resource languages, which are often overlooked in favour of English. Current cross-lingual ABSA approaches focus on limited, less complex tasks and often rely on external translation tools. This paper introduces a novel approach using constrained decoding with sequence-to-sequence models, eliminating the need for unreliable translation tools and improving cross-lingual performance by 5\% on average for the most complex task. The proposed method also supports multi-tasking, which enables solving multiple ABSA tasks with a single model, with constrained decoding boosting results by more than 10\%.
We evaluate our approach across seven languages and six ABSA tasks, surpassing state-of-the-art methods and setting new benchmarks for previously unexplored tasks. Additionally, we assess large language models (LLMs) in zero-shot, few-shot, and fine-tuning scenarios. While LLMs perform poorly in zero-shot and few-shot settings, fine-tuning achieves competitive results compared to smaller multilingual models, albeit at the cost of longer training and inference times.
We provide practical recommendations for real-world applications, enhancing the understanding of cross-lingual ABSA methodologies. This study offers valuable insights into the strengths and limitations of cross-lingual ABSA approaches, advancing the state-of-the-art in this challenging research domain.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:07:53 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Šmíd",
"Jakub",
""
],
[
"Přibáň",
"Pavel",
""
],
[
"Král",
"Pavel",
""
]
] |
2508.10370
|
Jiyong Kim
|
Jiyong Kim, Jaeho Lee, Jiahao Lin, Alish Kanani, Miao Sun, Umit Y. Ogras, and Jaehyun Park
|
eMamba: Efficient Acceleration Framework for Mamba Models in Edge Computing
|
Paper accepted at ESWEEK 2025 (CODES+ISSS) conference
| null | null | null |
cs.LG cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
State Space Model (SSM)-based machine learning architectures have recently gained significant attention for processing sequential data. Mamba, a recent sequence-to-sequence SSM, offers competitive accuracy with superior computational efficiency compared to state-of-the-art transformer models. While this advantage makes Mamba particularly promising for resource-constrained edge devices, no hardware acceleration frameworks are currently optimized for deploying it in such environments. This paper presents eMamba, a comprehensive end-to-end hardware acceleration framework explicitly designed for deploying Mamba models on edge platforms. eMamba maximizes computational efficiency by replacing complex normalization layers with lightweight hardware-aware alternatives and approximating expensive operations, such as SiLU activation and exponentiation, considering the target applications. Then, it performs an approximation-aware neural architecture search (NAS) to tune the learnable parameters used during approximation. Evaluations with Fashion-MNIST, CIFAR-10, and MARS, an open-source human pose estimation dataset, show eMamba achieves comparable accuracy to state-of-the-art techniques using 1.63-19.9$\times$ fewer parameters. In addition, it generalizes well to large-scale natural language tasks, demonstrating stable perplexity across varying sequence lengths on the WikiText2 dataset. We also quantize and implement the entire eMamba pipeline on an AMD ZCU102 FPGA and ASIC using GlobalFoundries (GF) 22 nm technology. Experimental results show 4.95-5.62$\times$ lower latency and 2.22-9.95$\times$ higher throughput, with 4.77$\times$ smaller area, 9.84$\times$ lower power, and 48.6$\times$ lower energy consumption than baseline solutions while maintaining competitive accuracy.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:08:05 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Kim",
"Jiyong",
""
],
[
"Lee",
"Jaeho",
""
],
[
"Lin",
"Jiahao",
""
],
[
"Kanani",
"Alish",
""
],
[
"Sun",
"Miao",
""
],
[
"Ogras",
"Umit Y.",
""
],
[
"Park",
"Jaehyun",
""
]
] |
2508.10371
|
Wenqi Zheng
|
Wenqi Zheng and Yutaka Arakawa
|
Few-shot Vision-based Human Activity Recognition with MLLM-based Visual Reinforcement Learning
| null | null | null | null |
cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Reinforcement learning in large reasoning models enables learning from feedback on their outputs, making it particularly valuable in scenarios where fine-tuning data is limited. However, its application in multi-modal human activity recognition (HAR) domains remains largely underexplored. Our work extends reinforcement learning to the human activity recognition domain with multimodal large language models. By incorporating visual reinforcement learning in the training process, the model's generalization ability on few-shot recognition can be greatly improved. Additionally, visual reinforcement learning can enhance the model's reasoning ability and enable explainable analysis in the inference stage. We name our few-shot human activity recognition method with visual reinforcement learning FAVOR. Specifically, our approach first utilizes a multimodal large language model (MLLM) to generate multiple candidate responses for the human activity image, each containing reasoning traces and final answers. These responses are then evaluated using reward functions, and the MLLM model is subsequently optimized using the Group Relative Policy Optimization (GRPO) algorithm. In this way, the MLLM model can be adapted to human activity recognition with only a few samples. Extensive experiments on four human activity recognition datasets and five different settings demonstrate the superiority of the proposed method.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:08:24 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zheng",
"Wenqi",
""
],
[
"Arakawa",
"Yutaka",
""
]
] |
2508.10372
|
Zitong Fang
|
Zitong Fang, Yejian Lyu, Ziming Yu, and Chong Han
|
Environment Reconstruction in Terahertz Monostatic Sensing: Joint Millimeter-level Geometry Mapping and Material Identification
| null | null | null | null |
eess.SP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Terahertz (THz) integrated sensing and communication (ISAC) offers high-speed communication alongside precise environmental sensing. This paper presents a computationally efficient framework for THz-based environment reconstruction by integrating connected component analysis (CCA)-assisted multipath component (MPC) estimation with a sliding-window refinement strategy. To start with, a monostatic sensing experiment is conducted in an indoor scenario using a vector network analyzer (VNA)-based sounder operating from 290 to 310 GHz. On one hand, as for geometry mapping, a CCA-based region search is employed to accelerate parameter extraction, significantly reducing the search space for space-alternating generalized expectation-maximization (SAGE)-based estimation and achieving an 8.4 times acceleration, while preserving resolution. Further analysis of the connected component structure enables the identification of indoor features such as flat walls and corners. A sliding-window refinement applied to the identified regions improves geometric mapping, achieving the mean distance error of 4.9 mm, which is one order of magnitude better than the literature. On the other hand, the deterministic and stochastic components of the monostatic channel are classified through reflection loss analysis. Then, material identification is performed by looking up the reflection loss in a THz time-domain spectroscopy (THz-TDS) database, which comprises over 200 materials across a 0-6 THz range. Experimental results validate millimeter-level accuracy in geometry mapping and reliable material classification, enhancing the environmental awareness capabilities of THz ISAC systems.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:09:28 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Fang",
"Zitong",
""
],
[
"Lyu",
"Yejian",
""
],
[
"Yu",
"Ziming",
""
],
[
"Han",
"Chong",
""
]
] |
2508.10373
|
Yingfan Liu
|
Yingfan Liu, Yandi Zhang, Jiadong Xie, Hui Li, Jeffrey Xu Yu, Jiangtao Cui
|
Privacy-Preserving Approximate Nearest Neighbor Search on High-Dimensional Data
|
This paper has been accepted by ICDE 2025
| null |
10.1109/ICDE65448.2025.00226
| null |
cs.DB
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In the era of cloud computing and AI, data owners outsource ubiquitous vectors to the cloud, which furnish approximate $k$-nearest neighbors ($k$-ANNS) services to users. To protect data privacy against the untrusted server, privacy-preserving $k$-ANNS (PP-ANNS) on vectors has been a fundamental and urgent problem. However, existing PP-ANNS solutions fall short of meeting the requirements of data privacy, efficiency, accuracy, and minimal user involvement concurrently. To tackle this challenge, we introduce a novel solution that primarily executes PP-ANNS on a single cloud server to avoid the heavy communication overhead between the cloud and the user. To ensure data privacy, we introduce a novel encryption method named distance comparison encryption, facilitating secure, efficient, and exact distance comparisons. To optimize the trade-off between data privacy and search performance, we design a privacy-preserving index that combines the state-of-the-art $k$-ANNS method with an approximate distance computation method. Then, we devise a search method using a filter-and-refine strategy based on the index. Moreover, we provide the security analysis of our solution and conduct extensive experiments to demonstrate its superiority over existing solutions. Based on our experimental results, our method accelerates PP-ANNS by up to 3 orders of magnitude compared to state-of-the-art methods, while not compromising the accuracy.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:09:38 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Liu",
"Yingfan",
""
],
[
"Zhang",
"Yandi",
""
],
[
"Xie",
"Jiadong",
""
],
[
"Li",
"Hui",
""
],
[
"Yu",
"Jeffrey Xu",
""
],
[
"Cui",
"Jiangtao",
""
]
] |
2508.10374
|
Michael Kuhlmann
|
Michael Kuhlmann, Fritz Seebauer, Petra Wagner, Reinhold Haeb-Umbach
|
Towards Frame-level Quality Predictions of Synthetic Speech
|
Accepted at Interspeech 2025
| null | null | null |
eess.AS
|
http://creativecommons.org/licenses/by/4.0/
|
While automatic subjective speech quality assessment has witnessed much progress, an open question is whether an automatic quality assessment at frame resolution is possible. This would be highly desirable, as it adds explainability to the assessment of speech synthesis systems. Here, we take first steps towards this goal by identifying issues of existing quality predictors that prevent sensible frame-level prediction. Further, we define criteria that a frame-level predictor should fulfill. We also suggest a chunk-based processing that avoids the impact of a localized distortion on the score of neighboring frames. Finally, we measure in experiments with localized artificial distortions the localization performance of a set of frame-level quality predictors and show that they can outperform detection performance of human annotations obtained from a crowd-sourced perception experiment.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:11:00 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Kuhlmann",
"Michael",
""
],
[
"Seebauer",
"Fritz",
""
],
[
"Wagner",
"Petra",
""
],
[
"Haeb-Umbach",
"Reinhold",
""
]
] |
2508.10375
|
Alja\v{z} Zalar
|
Lawrence Fialkow, Alja\v{z} Zalar
|
The pure $Y=X^{d}$ truncated moment problem
|
45 pages
| null | null | null |
math.FA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Let $\beta \equiv\beta^{(2n)}$ be a real bivariate sequence of degree $2n$. We study the existence of representing measures for $\beta$ supported in the planar curve $y=x^{d}$ in the case where the associated moment matrix $M_{n}(\beta)$ is $(y-x^{d})$-pure. In Section 2 we provide a general necessary and sufficient condition for representing measures in terms of positive semidefiniteness and recursive generation of the associated core matrix. In particular, if the core matrix is positive definite, then the core variety for $\beta$, i.e., the union of supports of all finitely atomic representing measures, is the whole curve $y = x^d$. In later sections, we provide various other concrete necessary or sufficient conditions for measures. For $d=3$, we provide a core-variety proof of the result of the first-named author characterizing the existence of representing measures. For $d \geq 4$ we develop a sufficient condition for the core variety to be finite or empty. For $d=4$, we adapt the technique of the second-named author, involving positive completions of partially-defined Hankel matrices, to provide necessary and sufficient numerical conditions for representing measures. We conclude with an example showing that in the $d=4$ case, the core variety can be finite, with a unique representing measure, which cannot occur for $d< 4$.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:11:18 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Fialkow",
"Lawrence",
""
],
[
"Zalar",
"Aljaž",
""
]
] |
2508.10376
|
Hangyu Xu
|
Yu Chen, Zihan Tan, Hangyu Xu
|
Lower Bounds on Tree Covers
| null | null | null | null |
cs.DS math.CO
|
http://creativecommons.org/licenses/by/4.0/
|
Given an $n$-point metric space $(X,d_X)$, a tree cover $\mathcal{T}$ is a set of $|\mathcal{T}|=k$ trees on $X$ such that every pair of vertices in $X$ has a low-distortion path in one of the trees in $\mathcal{T}$. Tree covers have been playing a crucial role in graph algorithms for decades, and the research focus is the construction of tree covers with small size $k$ and distortion.
When $k=1$, the best distortion is known to be $\Theta(n)$. For a constant $k\ge 2$, the best distortion upper bound is $\tilde O(n^{\frac 1 k})$ and the strongest lower bound is $\Omega(\log_k n)$, leaving a gap to be closed. In this paper, we improve the lower bound to $\Omega(n^{\frac{1}{2^{k-1}}})$.
Our proof is a novel analysis on a structurally simple grid-like graph, which utilizes some combinatorial fixed-point theorems. We believe that they will prove useful for analyzing other tree-like data structures as well.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:11:37 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Chen",
"Yu",
""
],
[
"Tan",
"Zihan",
""
],
[
"Xu",
"Hangyu",
""
]
] |
2508.10377
|
Michael Weiss
|
Michael Weiss, Robert Rosenbach, Christian Eggenberger
|
Clicks Versus Conversion: Choosing a Recommender's Training Objective in E-Commerce
| null | null | null | null |
cs.IR cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Ranking product recommendations to optimize for a high click-through rate (CTR) or for high conversion, such as add-to-cart rate (ACR) and Order-Submit-Rate (OSR, view-to-purchase conversion) are standard practices in e-commerce. Optimizing for CTR appears like a straightforward choice: Training data (i.e., click data) are simple to collect and often available in large quantities. Additionally, CTR is used far beyond e-commerce, making it a generalist, easily implemented option. ACR and OSR, on the other hand, are more directly linked to a shop's business goals, such as the Gross Merchandise Value (GMV). In this paper, we compare the effects of using either of these objectives using an online A/B test. Among our key findings, we demonstrate that in our shops, optimizing for OSR produces a GMV uplift more than five times larger than when optimizing for CTR, without sacrificing new product discovery. Our results also provide insights into the different feature importances for each of the objectives.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:15:23 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Weiss",
"Michael",
""
],
[
"Rosenbach",
"Robert",
""
],
[
"Eggenberger",
"Christian",
""
]
] |
2508.10378
|
Chen Yu
|
Yu Chen, Shu Miao, Chunyu Wu, Jingsong Mu, Bo OuYang, and Xiang Li
|
A Semantic-Aware Framework for Safe and Intent-Integrative Assistance in Upper-Limb Exoskeletons
| null | null | null | null |
cs.RO
|
http://creativecommons.org/licenses/by/4.0/
|
Upper-limb exoskeletons are primarily designed to provide assistive support by accurately interpreting and responding to human intentions. In home-care scenarios, exoskeletons are expected to adapt their assistive configurations based on the semantic information of the task, adjusting appropriately in accordance with the nature of the object being manipulated. However, existing solutions often lack the ability to understand task semantics or collaboratively plan actions with the user, limiting their generalizability. To address this challenge, this paper introduces a semantic-aware framework that integrates large language models into the task planning framework, enabling the delivery of safe and intent-integrative assistance. The proposed approach begins with the exoskeleton operating in transparent mode to capture the wearer's intent during object grasping. Once semantic information is extracted from the task description, the system automatically configures appropriate assistive parameters. In addition, a diffusion-based anomaly detector is used to continuously monitor the state of human-robot interaction and trigger real-time replanning in response to detected anomalies. During task execution, online trajectory refinement and impedance control are used to ensure safety and regulate human-robot interaction. Experimental results demonstrate that the proposed method effectively aligns with the wearer's cognition, adapts to semantically varying tasks, and responds reliably to anomalies.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:18:20 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Chen",
"Yu",
""
],
[
"Miao",
"Shu",
""
],
[
"Wu",
"Chunyu",
""
],
[
"Mu",
"Jingsong",
""
],
[
"OuYang",
"Bo",
""
],
[
"Li",
"Xiang",
""
]
] |
2508.10379
|
Lesley De Putter
|
Lesley G.A. de Putter and Marloes M.H.G. Hendrickx
|
Teachers of bachelors lab courses collaborating to promote open inquiry: a case study
|
Study on a learning community on the request of the learning community members themselves. 16 pages, 1 figure, 4 tables
| null | null | null |
physics.ed-ph
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
A group of university bachelors teachers of open inquiry lab courses in science subjects collaborated in a Networked Faculty Learning Community (NFLC) with the final goal to share their course materials publicly. In this paper we describe their collaborative process, communication on the topic and their road to a common definition of open inquiry as a didactical method. The study emerged organically during the collaboration of this NFLC as the group increasingly recognized the value of documenting their endeavours, allowing others to learn and grow as they did. We can conclude that learning takes place in such a community, even though professional development was not the NFLC s intention.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:18:48 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"de Putter",
"Lesley G. A.",
""
],
[
"Hendrickx",
"Marloes M. H. G.",
""
]
] |
2508.10380
|
Qisheng Yu
|
Qisheng Yu, Boyu Liu, Hongjun Xiang, Shi Liu
|
Type-I Multiferroic VHfO$_4$ with Strain-Switchable Magnetic Orders and Magnetoelectric Coupling
| null | null | null | null |
cond-mat.mtrl-sci physics.comp-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Motivated by the complementary properties of vanadium-based ferromagnets and HfO$_2$-based ferroelectrics, we propose a novel multiferroic oxide, VHfO$_4$, through 50\% Hf$^{4+}$ substitution with V$^{4+}$ in the ferroelectric $Pca2_1$ phase of HfO$_2$. First-principles DFT calculations reveal that the $Pca2_1$-like VHfO$_4$ phase exhibits dynamic stability and concurrent ferroic orders: robust ferroelectric polarization comparable to HfO$_2$ and V-driven magnetism. Parallel tempering Monte Carlo simulations identify an antiferromagnetic ground state, while strain engineering enables tunable magnetoelectric coupling. Biaxial in-plane strain induces four magnetic states: intralayer FM/interlayer AFM, intralayer AFM/interlayer FM, spiral-like non-collinear order, and discrete alternating spin alignment. Critically, $c$-axis strain modulates magnetic energy landscapes, demonstrating electromechanical control of magnetism. This work establishes VHfO$_4$ as a Type-I multiferroic with coexisting atomic-scale ferroic origins and strain-tunable cross-coupling, offering a platform for voltage-controlled spintronics devices.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:22:49 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Yu",
"Qisheng",
""
],
[
"Liu",
"Boyu",
""
],
[
"Xiang",
"Hongjun",
""
],
[
"Liu",
"Shi",
""
]
] |
2508.10381
|
Paul-Julius Hillmann
|
Paul-Julius Hillmann, Stephan A. Fahrenkrog-Petersen, Jan Mendling
|
Cross-Organizational Analysis of Parliamentary Processes: A Case Study
|
Accepted to ICPM 2025 (7th International Conference on Process Mining)
| null | null | null |
cs.DB
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Process Mining has been widely adopted by businesses and has been shown to help organizations analyze and optimize their processes. However, so far, little attention has gone into the cross-organizational comparison of processes, since many companies are hesitant to share their data. In this paper, we explore the processes of German state parliaments that are often legally required to share their data and run the same type of processes for different geographical regions. This paper is the first attempt to apply process mining to parliamentary processes and, therefore, contributes toward a novel interdisciplinary research area that combines political science and process mining. In our case study, we analyze legislative processes of three German state parliaments and generate insights into their differences and best practices. We provide a discussion of the relevance of our results that are based on knowledge exchange with a political scientist and a domain expert from the German federal parliament.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:25:59 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Hillmann",
"Paul-Julius",
""
],
[
"Fahrenkrog-Petersen",
"Stephan A.",
""
],
[
"Mendling",
"Jan",
""
]
] |
2508.10382
|
Hyundo Lee
|
Hyundo Lee, Suhyung Choi, Byoung-Tak Zhang, Inwoo Hwang
|
Towards Spatially Consistent Image Generation: On Incorporating Intrinsic Scene Properties into Diffusion Models
| null | null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Image generation models trained on large datasets can synthesize high-quality images but often produce spatially inconsistent and distorted images due to limited information about the underlying structures and spatial layouts. In this work, we leverage intrinsic scene properties (e.g., depth, segmentation maps) that provide rich information about the underlying scene, unlike prior approaches that solely rely on image-text pairs or use intrinsics as conditional inputs. Our approach aims to co-generate both images and their corresponding intrinsics, enabling the model to implicitly capture the underlying scene structure and generate more spatially consistent and realistic images. Specifically, we first extract rich intrinsic scene properties from a large image dataset with pre-trained estimators, eliminating the need for additional scene information or explicit 3D representations. We then aggregate various intrinsic scene properties into a single latent variable using an autoencoder. Building upon pre-trained large-scale Latent Diffusion Models (LDMs), our method simultaneously denoises the image and intrinsic domains by carefully sharing mutual information so that the image and intrinsic reflect each other without degrading image quality. Experimental results demonstrate that our method corrects spatial inconsistencies and produces a more natural layout of scenes while maintaining the fidelity and textual alignment of the base model (e.g., Stable Diffusion).
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:26:36 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Lee",
"Hyundo",
""
],
[
"Choi",
"Suhyung",
""
],
[
"Zhang",
"Byoung-Tak",
""
],
[
"Hwang",
"Inwoo",
""
]
] |
2508.10383
|
Yechan Kim
|
Yechan Kim, Dongho Yoon, Younkwan Lee, Unse Fatima, Hong Kook Kim, Songjae Lee, Sanga Park, Jeong Ho Park, Seonjong Kang, Moongu Jeon
|
Unlocking Robust Semantic Segmentation Performance via Label-only Elastic Deformations against Implicit Label Noise
| null | null | null | null |
cs.CV cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
While previous studies on image segmentation focus on handling severe (or explicit) label noise, real-world datasets also exhibit subtle (or implicit) label imperfections. These arise from inherent challenges, such as ambiguous object boundaries and annotator variability. Although not explicitly present, such mild and latent noise can still impair model performance. Typical data augmentation methods, which apply identical transformations to the image and its label, risk amplifying these subtle imperfections and limiting the model's generalization capacity. In this paper, we introduce NSegment+, a novel augmentation framework that decouples image and label transformations to address such realistic noise for semantic segmentation. By introducing controlled elastic deformations only to segmentation labels while preserving the original images, our method encourages models to focus on learning robust representations of object structures despite minor label inconsistencies. Extensive experiments demonstrate that NSegment+ consistently improves performance, achieving mIoU gains of up to +2.29, +2.38, +1.75, and +3.39 in average on Vaihingen, LoveDA, Cityscapes, and PASCAL VOC, respectively-even without bells and whistles, highlighting the importance of addressing implicit label noise. These gains can be further amplified when combined with other training tricks, including CutMix and Label Smoothing.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:27:43 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Kim",
"Yechan",
""
],
[
"Yoon",
"Dongho",
""
],
[
"Lee",
"Younkwan",
""
],
[
"Fatima",
"Unse",
""
],
[
"Kim",
"Hong Kook",
""
],
[
"Lee",
"Songjae",
""
],
[
"Park",
"Sanga",
""
],
[
"Park",
"Jeong Ho",
""
],
[
"Kang",
"Seonjong",
""
],
[
"Jeon",
"Moongu",
""
]
] |
2508.10384
|
Ruijuan Xiao
|
Liqi Wang, Ruijuan Xiao, and Hong Li
|
EDIS: A Simulation Software for Dynamic Ion Intercalation/Deintercalation Processes in Electrode Materials
| null | null | null | null |
cond-mat.mtrl-sci
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
As the core determinant of lithium-ion battery performance, electrode materials play a crucial role in defining the battery's capacity, cycling stability, and durability. During charging and discharging, electrode materials undergo complex ion intercalation and deintercalation processes, accompanied by defect formation and structural evolution. However, the microscopic mechanisms underlying processes such as cation disordering, lattice oxygen loss, and stage structure formation phenomena are still not fully understood. To address these challenges, we have developed the Electrode Dynamic Ion Intercalation/Deintercalation Simulator (EDIS), a software platform designed to simulate the dynamic processes of ion intercalation and deintercalation in electrode materials. Leveraging high-precision machine learning potentials, EDIS can efficiently model structural evolution and lithium-ion diffusion behavior under various states of charge and discharge, achieving accuracy approaching that of quantum mechanical methods in relevant chemical spaces. The software supports quantitative analysis of how variations in lithium-ion concentration and distribution affect lithium-ion transport properties, enables evaluation of the impact of structural defects, and allows for tracking of both structural evolution and transport characteristics during continuous cycling. EDIS is versatile and can be extended to sodium-ion batteries and related systems. By enabling in-depth analysis of these microscopic processes, EDIS provides a robust theoretical tool for mechanistic studies and the rational design of high-performance electrode materials for next-generation lithium-ion batteries.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:37:04 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Wang",
"Liqi",
""
],
[
"Xiao",
"Ruijuan",
""
],
[
"Li",
"Hong",
""
]
] |
2508.10385
|
Jongjun M. Lee
|
Jongjun M. Lee
|
Universal intrinsic orbital dynamics from Berry curvature in electronic two-band systems
|
8 pages, 2 figures, Accepted version to Physical Review B
| null | null | null |
cond-mat.mes-hall
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The geometric structure of quantum states plays a fundamental role in determining the intrinsic dynamics of electrons in solids. In this work, we study the geometric origin of orbital angular momentum and its transport in a general two-band electronic system. Without assuming any symmetry or dimensional constraints, we show that the orbital Berry curvature, which governs the orbital Hall effect, can be universally expressed as the product of the band energy and the square of the Berry curvature. This highlights the central role of Berry curvature in engineering orbital Hall responses. We also discuss the applicability of our framework by analyzing a realistic model. Our findings underscore the geometric universality of itinerant intrinsic orbital dynamics.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:39:41 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Lee",
"Jongjun M.",
""
]
] |
2508.10386
|
Tailong Zhang
|
Tailong Zhang, Benjamin Wilcock, Sushuang Ma, Giovanna Tinetti, Lawrence Bradley, Marcell Tessenyi and Jonathan Tennyson
|
Atmospheric Characterisation with the Twinkle Space Telescope Following Advances from JWST Observations
| null | null | null | null |
astro-ph.IM astro-ph.EP
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
The Twinkle Space Telescope is a satellite designed for spectroscopic observations of a wide range of extrasolar and solar system objects. Equipped with a 0.45 m diameter telescope and a spectrometer covering from 0.5 to 4.5 microns simultaneously, Twinkle will be launched in a sun-synchronous, low-Earth orbit, and it is expected to operate for seven years. Twinkle is developed, managed and operated by Blue Skies Space (BSSL), a space science data company whose vision is to accelerate and expand the availability of new, high-quality datasets to researchers worldwide, complementing the space observatories delivered by government space agencies.
Over its lifetime, Twinkle will conduct large-scale survey programs. The scientific objectives and observational strategy of these surveys are defined by researchers who join the Science Team.
Leveraging advances made possible by recent observations with the James Webb Space Telescope, we present here updated simulations evaluating Twinkle's observational capabilities in the context of exoplanet atmospheres. Through retrieval analyses of HD 209458 b, WASP-107 b, GJ 3470 b, and 55 Cnc e, we demonstrate how increasing observational investment enhances the retrieval of atmospheric parameters and molecular abundances. Our sensitivity study highlights Twinkle's capability to detect less abundant or less detectable molecules depending on the observing strategies adopted. This work provides practical guidance for developing targeted observational strategies to maximise Twinkle's scientific return.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:41:48 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Tailong",
""
],
[
"Wilcock",
"Benjamin",
""
],
[
"Ma",
"Sushuang",
""
],
[
"Tinetti",
"Giovanna",
""
],
[
"Bradley",
"Lawrence",
""
],
[
"Tessenyi",
"Marcell",
""
],
[
"Tennyson",
"Jonathan",
""
]
] |
2508.10387
|
Giusi Vaira
|
Giusi Vaira
|
Blow-up phenomena for a boundary Yamabe problem with umbilic boundary
| null | null | null | null |
math.AP
|
http://creativecommons.org/licenses/by/4.0/
|
We consider a linear perturbation of the classical geometric problem of prescribing the scalar and the boundary mean curvature problem in a Riemannian manifold with umbilic boundary provided the Weyl tensor is non-zero everywhere. We will deal with the case of negative scalar curvature showing the existence of a positive solutions when $n\geq 8$.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:43:13 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Vaira",
"Giusi",
""
]
] |
2508.10388
|
Liang Zhang
|
Liang Zhang, Takumi Doi, Yan Lyu, Tetsuo Hatsuda and Yu-Gang Ma
|
Probing Nucleon-$\Omega_{\rm ccc}$Interaction via Lattice QCD at Physical Quark Masses
|
8 pages, 6 figures and 3 tables
| null | null |
RIKEN-iTHEMS-Report-25
|
hep-lat hep-ph nucl-th
|
http://creativecommons.org/licenses/by/4.0/
|
We study the S-wave interactions between the nucleon ($N$) and the triply charmed Omega baryon ($\Omega_{\mathrm{ccc}}$) using (2+1)-flavor lattice QCD with a physical pion mass ($m_\pi \simeq 137.1$ MeV) on a lattice volume $\simeq (8.1~\mathrm{fm})^3$. The charm quark is implemented with a relativistic heavy-quark action at its physical mass. Employing the time-dependent HAL QCD method, the $N$-$\Omega_{\mathrm{ccc}}$ potentials in the spin-1 ($^3\mathrm{S}_1$) and spin-2 ($^5\mathrm{S}_2$) channels are extracted. In both channels, overall attraction is found with the scattering parameters, $a_0 = 0.56(0.13)\left(^{+0.26}_{-0.03}\right)$ fm and $r_{\mathrm{eff}} = 1.60(0.05)\left(^{+0.04}_{-0.12}\right)$ fm for the $^3\mathrm{S}_1$ channel, and
$a_0 = 0.38(0.12)\left(^{+0.25}_{-0.00}\right)$ fm and $r_{\mathrm{eff}} = 2.04(0.10)\left(^{+0.03}_{-0.22}\right)$ fm for the $^5\mathrm{S}_2$ channel, indicating the absence of a dibaryon bound state. The extracted potentials are further decomposed into spin-independent and spin-dependent components. The spin-independent potential is a dominant component and features a short-range attractive core and a long-range attractive tail, while the spin-dependent potential shows short-range attraction (repulsion) in the spin-1 (spin-2) channel. Qualitative comparisons with previous studies of the $N$-$J/\psi$ and $N$-$\Omega_{\rm{sss}}$ systems at $m_\pi \simeq 146$ MeV are provided, emphasizing the role of heavy-hadron chromo-polarizability arising from soft-gluon exchange between the nucleon and flavor-singlet hadrons. The charm quark mass dependence of the $N$-$\Omega_{\rm ccc}$ potential is investigated as well.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:44:11 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Liang",
""
],
[
"Doi",
"Takumi",
""
],
[
"Lyu",
"Yan",
""
],
[
"Hatsuda",
"Tetsuo",
""
],
[
"Ma",
"Yu-Gang",
""
]
] |
2508.10389
|
Wenlin Li
|
Wenlin Li, Xingli Li, Najmeh Eshaqi-Sani, Wen-zhao Zhang, and Jiong Cheng
|
Probing minimal observable length with dark modes in an optomechanical detector
| null | null | null | null |
quant-ph
|
http://creativecommons.org/licenses/by/4.0/
|
Several theories that attempt to unify quantum theory and gravitational theory assume that space has an observable limiting resolution related to the Planck length, denoted by $\sqrt{\beta_0}L_p$. Quantum mechanically, this concept derives a generalized uncertainty principle (GUP) and the corresponding modified commutator. The prediction and observation of GUP-induced new physics, as well as the quantitative measurement of the value of $\beta_0$, may provide substantial support for the establishment of quantum gravity theory. In this paper, we propose a comprehensive quantum framework for measuring GUP at low energy scales by utilizing the interference-induced bright-dark mode effect of oscillators in an optomechanical system. The nonlinearity induced by GUP will be amplified by the bright mode dynamics, and then be quantitatively read out by the noise spectrum of the dark mode. The measurement limit resolution of the scheme is not constrained by the quality factor of the oscillator. Under experimentally achievable parameters, the measurement resolution has been shown to reach $\beta_{\text{NL,lim}}=10^{-16.75}$, which is $10$ orders of magnitude lower than the electroweak level.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:44:31 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Li",
"Wenlin",
""
],
[
"Li",
"Xingli",
""
],
[
"Eshaqi-Sani",
"Najmeh",
""
],
[
"Zhang",
"Wen-zhao",
""
],
[
"Cheng",
"Jiong",
""
]
] |
2508.10390
|
Chiyu Zhang
|
Chiyu Zhang, Lu Zhou, Xiaogang Xu, Jiafei Wu, Liming Fang, Zhe Liu
|
Jailbreaking Commercial Black-Box LLMs with Explicitly Harmful Prompts
| null | null | null | null |
cs.CL cs.CR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Evaluating jailbreak attacks is challenging when prompts are not overtly harmful or fail to induce harmful outputs. Unfortunately, many existing red-teaming datasets contain such unsuitable prompts. To evaluate attacks accurately, these datasets need to be assessed and cleaned for maliciousness. However, existing malicious content detection methods rely on either manual annotation, which is labor-intensive, or large language models (LLMs), which have inconsistent accuracy in harmful types. To balance accuracy and efficiency, we propose a hybrid evaluation framework named MDH (Malicious content Detection based on LLMs with Human assistance) that combines LLM-based annotation with minimal human oversight, and apply it to dataset cleaning and detection of jailbroken responses. Furthermore, we find that well-crafted developer messages can significantly boost jailbreak success, leading us to propose two new strategies: D-Attack, which leverages context simulation, and DH-CoT, which incorporates hijacked chains of thought. The Codes, datasets, judgements, and detection results will be released in github repository: https://github.com/AlienZhang1996/DH-CoT.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:46:56 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Chiyu",
""
],
[
"Zhou",
"Lu",
""
],
[
"Xu",
"Xiaogang",
""
],
[
"Wu",
"Jiafei",
""
],
[
"Fang",
"Liming",
""
],
[
"Liu",
"Zhe",
""
]
] |
2508.10391
|
Yaoze Zhang
|
Yaoze Zhang, Rong Wu, Pinlong Cai, Xiaoman Wang, Guohang Yan, Song Mao, Ding Wang, Botian Shi
|
LeanRAG: Knowledge-Graph-Based Generation with Semantic Aggregation and Hierarchical Retrieval
| null | null | null | null |
cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Retrieval-Augmented Generation (RAG) plays a crucial role in grounding Large Language Models by leveraging external knowledge, whereas the effectiveness is often compromised by the retrieval of contextually flawed or incomplete information. To address this, knowledge graph-based RAG methods have evolved towards hierarchical structures, organizing knowledge into multi-level summaries. However, these approaches still suffer from two critical, unaddressed challenges: high-level conceptual summaries exist as disconnected ``semantic islands'', lacking the explicit relations needed for cross-community reasoning; and the retrieval process itself remains structurally unaware, often degenerating into an inefficient flat search that fails to exploit the graph's rich topology. To overcome these limitations, we introduce LeanRAG, a framework that features a deeply collaborative design combining knowledge aggregation and retrieval strategies. LeanRAG first employs a novel semantic aggregation algorithm that forms entity clusters and constructs new explicit relations among aggregation-level summaries, creating a fully navigable semantic network. Then, a bottom-up, structure-guided retrieval strategy anchors queries to the most relevant fine-grained entities and then systematically traverses the graph's semantic pathways to gather concise yet contextually comprehensive evidence sets. The LeanRAG can mitigate the substantial overhead associated with path retrieval on graphs and minimizes redundant information retrieval. Extensive experiments on four challenging QA benchmarks with different domains demonstrate that LeanRAG significantly outperforming existing methods in response quality while reducing 46\% retrieval redundancy. Code is available at: https://github.com/RaZzzyz/LeanRAG
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:47:18 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Yaoze",
""
],
[
"Wu",
"Rong",
""
],
[
"Cai",
"Pinlong",
""
],
[
"Wang",
"Xiaoman",
""
],
[
"Yan",
"Guohang",
""
],
[
"Mao",
"Song",
""
],
[
"Wang",
"Ding",
""
],
[
"Shi",
"Botian",
""
]
] |
2508.10392
|
Francesco Moscatelli
|
F. Moscatelli, A. Fondacci, R. Arcidiacono, M. Boscardin, N. Cartiglia, M. Centis Vignali, T. Croci, M. Ferrero, O. Hammad Ali, L. Lanteri, A. Losana, L. Menzio, V. Monaco, R. Mulargia, D. Passeri G. Paternoster, F. Siviero, V. Sola, A. Morozzi
|
Measurements and TCAD simulations of innovative RSD and DC-RSD LGAD devices for future 4D tracking
| null | null | null | null |
physics.ins-det
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This paper summarizes the beam test results obtained with a Resistive Silicon Detector (RSD) (also called AC-Low Gain Avalanche Diode, AC-LGAD) pixel array tested at the DESY beam test facility with a 5 GeV/c electron beam. Furthermore, it describes in detail the simulation results of DC-RSD, an evolution of the RSD design. The simulations campaign described in this paper has been instrumental in the definition of the structures implemented in the Fondazione Bruno Kessler FBK first DC-RSD production.
The RSD matrix used in this study is part of the second FBK RSD production, RSD2. The best position resolution reached in this test is sigma_x = 15 micron, about 3.4% of the pitch. DC-RSD LGAD, are an evolution of the AC-coupled design, eliminating the dielectric and using a DC-coupling to the electronics. The concept of DC-RSD has been finalized using full 3D Technology-CAD simulations of the sensor behavior. TCAD simulations are an excellent tool for designing this innovative class of detectors, enabling the evaluation of different technology options (e.g., the resistivity of the n+ layer, contact materials) and geometrical layouts (shape and distance of the read-out pads).
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:47:22 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Moscatelli",
"F.",
""
],
[
"Fondacci",
"A.",
""
],
[
"Arcidiacono",
"R.",
""
],
[
"Boscardin",
"M.",
""
],
[
"Cartiglia",
"N.",
""
],
[
"Vignali",
"M. Centis",
""
],
[
"Croci",
"T.",
""
],
[
"Ferrero",
"M.",
""
],
[
"Ali",
"O. Hammad",
""
],
[
"Lanteri",
"L.",
""
],
[
"Losana",
"A.",
""
],
[
"Menzio",
"L.",
""
],
[
"Monaco",
"V.",
""
],
[
"Mulargia",
"R.",
""
],
[
"Paternoster",
"D. Passeri G.",
""
],
[
"Siviero",
"F.",
""
],
[
"Sola",
"V.",
""
],
[
"Morozzi",
"A.",
""
]
] |
2508.10393
|
Liyun Zhang
|
Liyun Zhang, Jingcheng Ke, Shenli Fan, Xuanmeng Sha and Zheng Lian
|
A Unified Evaluation Framework for Multi-Annotator Tendency Learning
|
9 pages
| null | null | null |
cs.LG cs.MM
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent works have emerged in multi-annotator learning that shift focus from Consensus-oriented Learning (CoL), which aggregates multiple annotations into a single ground-truth prediction, to Individual Tendency Learning (ITL), which models annotator-specific labeling behavior patterns (i.e., tendency) to provide explanation analysis for understanding annotator decisions. However, no evaluation framework currently exists to assess whether ITL methods truly capture individual tendencies and provide meaningful behavioral explanations. To address this gap, we propose the first unified evaluation framework with two novel metrics: (1) Difference of Inter-annotator Consistency (DIC) quantifies how well models capture annotator tendencies by comparing predicted inter-annotator similarity structures with ground-truth; (2) Behavior Alignment Explainability (BAE) evaluates how well model explanations reflect annotator behavior and decision relevance by aligning explainability-derived with ground-truth labeling similarity structures via Multidimensional Scaling (MDS). Extensive experiments validate the effectiveness of our proposed evaluation framework.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:50:20 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Zhang",
"Liyun",
""
],
[
"Ke",
"Jingcheng",
""
],
[
"Fan",
"Shenli",
""
],
[
"Sha",
"Xuanmeng",
""
],
[
"Lian",
"Zheng",
""
]
] |
2508.10394
|
Kaitlin Ragosta
|
Kaitlin Ragosta
|
A marking graph for finite-type Artin groups
|
52 pages. Comments welcome
| null | null | null |
math.GR math.GT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Clean markings on surfaces were a key component in Masur and Minsky's hierarchy machinery, which proved to be a powerful tool in the study of mapping class groups. We construct a marking graph for irreducible finite-type Artin groups which is quasi-isometric to the group modulo its center, i.e., an element of $A_{\Gamma}/Z(A_{\Gamma})$ is determined up to finite error by its action on one of our markings. To construct this graph, we construct suitable collections of transverse parabolic subgroups which extend the maximal simplices of the complex of irreducible parabolic subgroups to analogues of clean markings, and we define natural analogues of elementary moves.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:51:41 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Ragosta",
"Kaitlin",
""
]
] |
2508.10395
|
Coleman Hooper
|
Aditya Tomar, Coleman Hooper, Minjae Lee, Haocheng Xi, Rishabh Tiwari, Wonjun Kang, Luca Manolache, Michael W. Mahoney, Kurt Keutzer, Amir Gholami
|
XQuant: Breaking the Memory Wall for LLM Inference with KV Cache Rematerialization
|
24 pages
| null | null | null |
cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Although LLM inference has emerged as a critical workload for many downstream applications, efficiently inferring LLMs is challenging due to the substantial memory footprint and bandwidth requirements. In parallel, compute capabilities have steadily outpaced both memory capacity and bandwidth over the last few decades, a trend that remains evident in modern GPU hardware and exacerbates the challenge of LLM inference. As such, new algorithms are emerging that trade increased computation for reduced memory operations. To that end, we present XQuant, which takes advantage of this trend, enabling an order-of-magnitude reduction in memory consumption through low-bit quantization with substantial accuracy benefits relative to state-of-the-art KV cache quantization methods. We accomplish this by quantizing and caching the layer input activations X, instead of using standard KV caching, and then rematerializing the Keys and Values on-the-fly during inference. This results in an immediate 2$\times$ memory savings compared to KV caching. By applying XQuant, we achieve up to $\sim 7.7\times$ memory savings with $<0.1$ perplexity degradation compared to the FP16 baseline. Furthermore, our approach leverages the fact that X values are similar across layers. Building on this observation, we introduce XQuant-CL, which exploits the cross-layer similarity in the X embeddings for extreme compression. Across different models, XQuant-CL attains up to 10$\times$ memory savings relative to the FP16 baseline with only 0.01 perplexity degradation, and 12.5$\times$ memory savings with only $0.1$ perplexity degradation. XQuant exploits the rapidly increasing compute capabilities of hardware platforms to eliminate the memory bottleneck, while surpassing state-of-the-art KV cache quantization methods and achieving near-FP16 accuracy across a wide range of models.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:52:38 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Tomar",
"Aditya",
""
],
[
"Hooper",
"Coleman",
""
],
[
"Lee",
"Minjae",
""
],
[
"Xi",
"Haocheng",
""
],
[
"Tiwari",
"Rishabh",
""
],
[
"Kang",
"Wonjun",
""
],
[
"Manolache",
"Luca",
""
],
[
"Mahoney",
"Michael W.",
""
],
[
"Keutzer",
"Kurt",
""
],
[
"Gholami",
"Amir",
""
]
] |
2508.10396
|
Vladimir Lisy
|
Vladimir Lisy, Jan Busa, Jana Tothova
|
Memory effects of a static magnetic field on Brownian motion and the question of the absence of classical magnetism
|
The work presented at the 18th Czech and Slovak Conference on Magnetism, High Tatras, Slovakia, July 7-11, 2025
| null | null | null |
cond-mat.soft
|
http://creativecommons.org/licenses/by/4.0/
|
The Bohr-van Leeuwen (BvL) theorem, stating the absence of classical magnetization in equilibrium, a fundamental result in the field of magnetic phenomena, was originally proved for an electron gas. In the present work, we deal with the problem of whether this theorem applies to particles undergoing a non-Markovian Brownian motion in a static magnetic field. We consider a charged Brownian particle (BP) immersed in a bath of neutral particles. Generalizing the Zwanzig-Caldeira-Legget theory to the presence of a static external magnetic field, we come to the equation of motion for the BP in the form of a generalized Langevin equation that accounts for memory effects in the dynamics of the system. By using its solutions for the displacement and velocity of the BP, we calculate the angular momentum for the Ornstein-Uhlenbeck thermal noise. At long times, when the system should reach equilibrium, this momentum and, consequently, the classical magnetic moment of the BP are nonzero, in contrast to the BvL theorem. With the help of analytical and precise numerical calculations for different sets of system parameters, a simple formula for the angular momentum has been deduced.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:53:44 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Lisy",
"Vladimir",
""
],
[
"Busa",
"Jan",
""
],
[
"Tothova",
"Jana",
""
]
] |
2508.10397
|
Xinghui Song
|
Haibin Sun, Xinghui Song
|
PQ-DAF: Pose-driven Quality-controlled Data Augmentation for Data-scarce Driver Distraction Detection
|
11 pages, 6 figures
| null | null | null |
cs.CV cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Driver distraction detection is essential for improving traffic safety and reducing road accidents. However, existing models often suffer from degraded generalization when deployed in real-world scenarios. This limitation primarily arises from the few-shot learning challenge caused by the high cost of data annotation in practical environments, as well as the substantial domain shift between training datasets and target deployment conditions. To address these issues, we propose a Pose-driven Quality-controlled Data Augmentation Framework (PQ-DAF) that leverages a vision-language model for sample filtering to cost-effectively expand training data and enhance cross-domain robustness. Specifically, we employ a Progressive Conditional Diffusion Model (PCDMs) to accurately capture key driver pose features and synthesize diverse training examples. A sample quality assessment module, built upon the CogVLM vision-language model, is then introduced to filter out low-quality synthetic samples based on a confidence threshold, ensuring the reliability of the augmented dataset. Extensive experiments demonstrate that PQ-DAF substantially improves performance in few-shot driver distraction detection, achieving significant gains in model generalization under data-scarce conditions.
|
[
{
"version": "v1",
"created": "Thu, 14 Aug 2025 06:54:28 GMT"
}
] | 2025-08-15T00:00:00 |
[
[
"Sun",
"Haibin",
""
],
[
"Song",
"Xinghui",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.