id
stringlengths 9
16
| submitter
stringlengths 1
64
⌀ | authors
stringlengths 3
62.8k
| title
stringlengths 1
382
| comments
stringlengths 1
1.15k
⌀ | journal-ref
stringlengths 1
557
⌀ | doi
stringlengths 7
153
⌀ | report-no
stringlengths 1
509
⌀ | categories
stringlengths 5
125
| license
stringclasses 9
values | abstract
stringlengths 6
6.09k
| versions
listlengths 1
187
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-06-27 00:00:00
| authors_parsed
sequencelengths 1
2.83k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.21030 | Tianxing Zhou | Zhou Tianxing, Wang Zhirui, Ao Haojia, Chen Guangyan, Xing Boyang, Cheng Jingwen, Yang Yi, Yue Yufeng | STEP Planner: Constructing cross-hierarchical subgoal tree as an embodied long-horizon task planner | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ability to perform reliable long-horizon task planning is crucial for deploying robots in real-world environments. However, directly employing Large Language Models (LLMs) as action sequence generators often results in low success rates due to their limited reasoning ability for long-horizon embodied tasks. In the STEP framework, we construct a subgoal tree through a pair of closed-loop models: a subgoal decomposition model and a leaf node termination model. Within this framework, we develop a hierarchical tree structure that spans from coarse to fine resolutions. The subgoal decomposition model leverages a foundation LLM to break down complex goals into manageable subgoals, thereby spanning the subgoal tree. The leaf node termination model provides real-time feedback based on environmental states, determining when to terminate the tree spanning and ensuring each leaf node can be directly converted into a primitive action. Experiments conducted in both the VirtualHome WAH-NL benchmark and on real robots demonstrate that STEP achieves long-horizon embodied task completion with success rates up to 34% (WAH-NL) and 25% (real robot) outperforming SOTA methods. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:10:02 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Tianxing",
"Zhou",
""
],
[
"Zhirui",
"Wang",
""
],
[
"Haojia",
"Ao",
""
],
[
"Guangyan",
"Chen",
""
],
[
"Boyang",
"Xing",
""
],
[
"Jingwen",
"Cheng",
""
],
[
"Yi",
"Yang",
""
],
[
"Yufeng",
"Yue",
""
]
] |
2506.21031 | Jatin Gupta | Jatin Gupta, Akhil Sharma, Saransh Singhania, Mohammad Adnan, Sakshi Deo, Ali Imam Abidi and Keshav Gupta | Large Language Models Acing Chartered Accountancy | Accepted for publication at MoStart 2025: International Conference on Digital Transformation in Education and Applications of Artificial Intelligence, Bosnia and Herzegovina, 2025 | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Advanced intelligent systems, particularly Large Language Models (LLMs), are significantly reshaping financial practices through advancements in Natural Language Processing (NLP). However, the extent to which these models effectively capture and apply domain-specific financial knowledge remains uncertain. Addressing a critical gap in the expansive Indian financial context, this paper introduces CA-Ben, a Chartered Accountancy benchmark specifically designed to evaluate the financial, legal, and quantitative reasoning capabilities of LLMs. CA-Ben comprises structured question-answer datasets derived from the rigorous examinations conducted by the Institute of Chartered Accountants of India (ICAI), spanning foundational, intermediate, and advanced CA curriculum stages. Six prominent LLMs i.e. GPT 4o, LLAMA 3.3 70B, LLAMA 3.1 405B, MISTRAL Large, Claude 3.5 Sonnet, and Microsoft Phi 4 were evaluated using standardized protocols. Results indicate variations in performance, with Claude 3.5 Sonnet and GPT-4o outperforming others, especially in conceptual and legal reasoning. Notable challenges emerged in numerical computations and legal interpretations. The findings emphasize the strengths and limitations of current LLMs, suggesting future improvements through hybrid reasoning and retrieval-augmented generation methods, particularly for quantitative analysis and accurate legal interpretation. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:10:37 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Gupta",
"Jatin",
""
],
[
"Sharma",
"Akhil",
""
],
[
"Singhania",
"Saransh",
""
],
[
"Adnan",
"Mohammad",
""
],
[
"Deo",
"Sakshi",
""
],
[
"Abidi",
"Ali Imam",
""
],
[
"Gupta",
"Keshav",
""
]
] |
2506.21032 | Jiangxia Cao | Shuo Yang, Jiangxia Cao, Haipeng Li, Yuqi Mao, Shuchao Pang | RecCoT: Enhancing Recommendation via Chain-of-Thought | Work in progress | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In real-world applications, users always interact with items in multiple aspects, such as through implicit binary feedback (e.g., clicks, dislikes, long views) and explicit feedback (e.g., comments, reviews). Modern recommendation systems (RecSys) learn user-item collaborative signals from these implicit feedback signals as a large-scale binary data-streaming, subsequently recommending other highly similar items based on users' personalized historical interactions. However, from this collaborative-connection perspective, the RecSys does not focus on the actual content of the items themselves but instead prioritizes higher-probability signals of behavioral co-occurrence among items. Consequently, under this binary learning paradigm, the RecSys struggles to understand why a user likes or dislikes certain items. To alleviate it, some works attempt to utilize the content-based reviews to capture the semantic knowledge to enhance recommender models. However, most of these methods focus on predicting the ratings of reviews, but do not provide a human-understandable explanation. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:14:42 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yang",
"Shuo",
""
],
[
"Cao",
"Jiangxia",
""
],
[
"Li",
"Haipeng",
""
],
[
"Mao",
"Yuqi",
""
],
[
"Pang",
"Shuchao",
""
]
] |
2506.21033 | Zhaojiacheng Zhou | Zhaojiacheng Zhou, Hongze Liu, Shijing Yuan, Hanning Zhang, Jiong Lou, Chentao Wu, Jie Li | BLOCKS: Blockchain-supported Cross-Silo Knowledge Sharing for Efficient LLM Services | null | null | null | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The hallucination problem of Large Language Models (LLMs) has increasingly drawn attention. Augmenting LLMs with external knowledge is a promising solution to address this issue. However, due to privacy and security concerns, a vast amount of downstream task-related knowledge remains dispersed and isolated across various "silos," making it difficult to access. To bridge this knowledge gap, we propose a blockchain-based external knowledge framework that coordinates multiple knowledge silos to provide reliable foundational knowledge for large model retrieval while ensuring data security. Technically, we distill knowledge from local data into prompts and execute transactions and records on the blockchain. Additionally, we introduce a reputation mechanism and cross-validation to ensure knowledge quality and provide incentives for participation. Furthermore, we design a query generation framework that provides a direct API interface for large model retrieval. To evaluate the performance of our proposed framework, we conducted extensive experiments on various knowledge sources. The results demonstrate that the proposed framework achieves efficient LLM service knowledge sharing in blockchain environments. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:16:33 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Zhou",
"Zhaojiacheng",
""
],
[
"Liu",
"Hongze",
""
],
[
"Yuan",
"Shijing",
""
],
[
"Zhang",
"Hanning",
""
],
[
"Lou",
"Jiong",
""
],
[
"Wu",
"Chentao",
""
],
[
"Li",
"Jie",
""
]
] |
2506.21034 | Wenzhou Lv | Wenzhou Lyu, Jialing Lin, Wenqi Ren, Ruihao Xia, Feng Qian, Yang Tang | DidSee: Diffusion-Based Depth Completion for Material-Agnostic Robotic Perception and Manipulation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Commercial RGB-D cameras often produce noisy, incomplete depth maps for non-Lambertian objects. Traditional depth completion methods struggle to generalize due to the limited diversity and scale of training data. Recent advances exploit visual priors from pre-trained text-to-image diffusion models to enhance generalization in dense prediction tasks. However, we find that biases arising from training-inference mismatches in the vanilla diffusion framework significantly impair depth completion performance. Additionally, the lack of distinct visual features in non-Lambertian regions further hinders precise prediction. To address these issues, we propose \textbf{DidSee}, a diffusion-based framework for depth completion on non-Lambertian objects. First, we integrate a rescaled noise scheduler enforcing a zero terminal signal-to-noise ratio to eliminate signal leakage bias. Second, we devise a noise-agnostic single-step training formulation to alleviate error accumulation caused by exposure bias and optimize the model with a task-specific loss. Finally, we incorporate a semantic enhancer that enables joint depth completion and semantic segmentation, distinguishing objects from backgrounds and yielding precise, fine-grained depth maps. DidSee achieves state-of-the-art performance on multiple benchmarks, demonstrates robust real-world generalization, and effectively improves downstream tasks such as category-level pose estimation and robotic grasping.Project page: https://wenzhoulyu.github.io/DidSee/ | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:18:42 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Lyu",
"Wenzhou",
""
],
[
"Lin",
"Jialing",
""
],
[
"Ren",
"Wenqi",
""
],
[
"Xia",
"Ruihao",
""
],
[
"Qian",
"Feng",
""
],
[
"Tang",
"Yang",
""
]
] |
2506.21035 | Haodong Lu | Haodong Lu, Chongyang Zhao, Jason Xue, Lina Yao, Kristen Moore, Dong Gong | Little By Little: Continual Learning via Self-Activated Sparse Mixture-of-Rank Adaptive Learning | Preprint | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Continual learning (CL) with large pre-trained models is challenged by catastrophic forgetting and task interference. Existing LoRA-based Mixture-of-Experts (MoE) approaches mitigate forgetting by assigning and freezing task-specific adapters, but suffer from interference, redundancy, and ambiguous routing due to coarse adapter-level selection. However, this design introduces three key challenges: 1) Interference: Activating full LoRA experts per input leads to subspace interference and prevents selective reuse of useful components across tasks. 2) Redundancy: Newly added experts often duplicate or contradict existing knowledge due to unnecessary activation of unrelated ranks and insufficient reuse of relevant ones. 3) Ambiguity: Overlapping features across tasks confuse the router, resulting in unstable expert assignments. As more experts accumulate, earlier task routing degrades, accelerating forgetting. We propose MoRA, a Mixture-of-Rank Adaptive learning approach with self-activated and sparse rank activation for CL. Unlike mixing multiple low-rank matrices, MoRA decomposes each rank-r update into r rank-1 components, each treated as an independent expert, enabling fine-grained mixture of rank-1 expert utilization while mitigating interference and redundancy. To avoid ambiguous routing, we propose that each rank-1 expert can infer its own relevance via intermediate activations. Coupled with our proposed rank pruning and activation budgets, MoRA adaptively selects a sparse mixture of ranks per input. We validate MoRA on continual learning tasks with CLIP and large language models (LLMs), analyzing both in-domain learning and out-of-domain forgetting/generalization during fine-tuning. MoRA shows significant effectiveness on enhancing CL with PTMs, and improving generalization while mitigating forgetting. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:19:05 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Lu",
"Haodong",
""
],
[
"Zhao",
"Chongyang",
""
],
[
"Xue",
"Jason",
""
],
[
"Yao",
"Lina",
""
],
[
"Moore",
"Kristen",
""
],
[
"Gong",
"Dong",
""
]
] |
2506.21036 | Fu Peng | Fu Peng, Meng Zhang and Ming Tang | An Information-Theoretic Analysis for Federated Learning under Concept Drift | null | null | null | null | cs.LG cs.DC | http://creativecommons.org/licenses/by/4.0/ | Recent studies in federated learning (FL) commonly train models on static datasets. However, real-world data often arrives as streams with shifting distributions, causing performance degradation known as concept drift. This paper analyzes FL performance under concept drift using information theory and proposes an algorithm to mitigate the performance degradation. We model concept drift as a Markov chain and introduce the \emph{Stationary Generalization Error} to assess a model's capability to capture characteristics of future unseen data. Its upper bound is derived using KL divergence and mutual information. We study three drift patterns (periodic, gradual, and random) and their impact on FL performance. Inspired by this, we propose an algorithm that regularizes the empirical risk minimization approach with KL divergence and mutual information, thereby enhancing long-term performance. We also explore the performance-cost tradeoff by identifying a Pareto front. To validate our approach, we build an FL testbed using Raspberry Pi4 devices. Experimental results corroborate with theoretical findings, confirming that drift patterns significantly affect performance. Our method consistently outperforms existing approaches for these three patterns, demonstrating its effectiveness in adapting concept drift in FL. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:25:15 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Peng",
"Fu",
""
],
[
"Zhang",
"Meng",
""
],
[
"Tang",
"Ming",
""
]
] |
2506.21037 | Suorong Yang | Suorong Yang, Peijia Li, Furao Shen, Jian Zhao | RL-Selector: Reinforcement Learning-Guided Data Selection via Redundancy Assessment | ICCV 2025 | ICCV 2025 | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Modern deep architectures often rely on large-scale datasets, but training on these datasets incurs high computational and storage overhead. Real-world datasets often contain substantial redundancies, prompting the need for more data-efficient training paradigms. Data selection has shown promise to mitigate redundancy by identifying the most representative samples, thereby reducing training costs without compromising performance. Existing methods typically rely on static scoring metrics or pretrained models, overlooking the combined effect of selected samples and their evolving dynamics during training. We introduce the concept of epsilon-sample cover, which quantifies sample redundancy based on inter-sample relationships, capturing the intrinsic structure of the dataset. Based on this, we reformulate data selection as a reinforcement learning (RL) process and propose RL-Selector, where a lightweight RL agent optimizes the selection policy by leveraging epsilon-sample cover derived from evolving dataset distribution as a reward signal. Extensive experiments across benchmark datasets and diverse architectures demonstrate that our method consistently outperforms existing state-of-the-art baselines. Models trained with our selected datasets show enhanced generalization performance with improved training efficiency. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:28:56 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yang",
"Suorong",
""
],
[
"Li",
"Peijia",
""
],
[
"Shen",
"Furao",
""
],
[
"Zhao",
"Jian",
""
]
] |
2506.21038 | Wei Kong | Tong Jiang (1,2,4), Han Chen (1,2,4), Yubo Yuan (1,2,4), Xiang Xu (1,2,4), Junwei Cao (1,2,4), Hao Wang (2,4), Xuechun Sun (1,2,4), Junshuai Li (2,4), Yaqing Ma (1,2,4), Huaze Zhu (2,4), Wenbin Li (2,4), Wei Kong (2,3,4,5) ((1) School of Materials Science and Engineering, Zhejiang University, Hangzhou, China, (2) Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China, (3) Research Center for Industries of the Future, Westlake University, Hangzhou, Zhejiang, China, (4) School of Engineering, Westlake University, Hangzhou, Zhejiang, China, (5) Westlake Institute for Optoelectronics, Hangzhou, China) | Ferroelectricity in 6 Angstrom-Thick Two-dimensional Ga$_2$O$_3$ | 38 pages, 13 figures | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Atomic-scale ferroelectric thin films hold great promise for high-density, low-power applications but face stability and voltage scaling challenges at extreme thinness. Here, we demonstrate ferroelectricity in single-crystalline two-dimensional (2D) Ga$_2$O$_3$, an ultra-wide-bandgap semiconductor, at just 6 angstrom thickness, exhibiting exceptional retention and thermal stability. We show that epitaxial beta-Ga$_2$O$_3$ can be exfoliated down to a half-unit cell thickness via a self-limiting mechanism, enabling a biaxial strain-induced phase transition into a novel ferroelectric layered structure. Strain modulation enables the reduction of polarization switching voltage to 0.8 V, meeting CMOS voltage scaling requirements. Theoretical calculations reveal that switching is driven by covalent bond reconstruction, effectively countering depolarization and enhancing stability. Additionally, we integrate ferroelectric 2D Ga$_2$O$_3$ onto silicon using a low-temperature, back-end-of-line-compatible process. This work advances the exploration of sub-nanometer ferroelectrics, paving the way for high-density, low-power, non-volatile applications seamlessly integrated with advanced silicon technology. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:30:29 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Jiang",
"Tong",
"",
"School of Materials Science and Engineering, Zhejiang University, Hangzhou, China",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Chen",
"Han",
"",
"School of Materials Science and Engineering, Zhejiang University, Hangzhou, China",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Yuan",
"Yubo",
"",
"School of Materials Science and Engineering, Zhejiang University, Hangzhou, China",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Xu",
"Xiang",
"",
"School of Materials Science and Engineering, Zhejiang University, Hangzhou, China",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Cao",
"Junwei",
"",
"School of Materials Science and Engineering, Zhejiang University, Hangzhou, China",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Wang",
"Hao",
"",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Sun",
"Xuechun",
"",
"School of Materials Science and Engineering, Zhejiang University, Hangzhou, China",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Li",
"Junshuai",
"",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Ma",
"Yaqing",
"",
"School of Materials Science and Engineering, Zhejiang University, Hangzhou, China",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Zhu",
"Huaze",
"",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Li",
"Wenbin",
"",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China"
],
[
"Kong",
"Wei",
"",
"Key Laboratory of 3D Micro/Nano Fabrication and Characterization of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, China",
"Research Center for Industries of the Future, Westlake University, Hangzhou, Zhejiang, China",
"School of Engineering, Westlake University, Hangzhou, Zhejiang, China",
"Westlake Institute for Optoelectronics, Hangzhou, China"
]
] |
2506.21039 | Seungyul Han | Jaebak Hwang, Sanghyeon Lee, Jeongmo Kim, Seungyul Han | Strict Subgoal Execution: Reliable Long-Horizon Planning in Hierarchical Reinforcement Learning | 9 technical page followed by references and appendix | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Long-horizon goal-conditioned tasks pose fundamental challenges for reinforcement learning (RL), particularly when goals are distant and rewards are sparse. While hierarchical and graph-based methods offer partial solutions, they often suffer from subgoal infeasibility and inefficient planning. We introduce Strict Subgoal Execution (SSE), a graph-based hierarchical RL framework that enforces single-step subgoal reachability by structurally constraining high-level decision-making. To enhance exploration, SSE employs a decoupled exploration policy that systematically traverses underexplored regions of the goal space. Furthermore, a failure-aware path refinement, which refines graph-based planning by dynamically adjusting edge costs according to observed low-level success rates, thereby improving subgoal reliability. Experimental results across diverse long-horizon benchmarks demonstrate that SSE consistently outperforms existing goal-conditioned RL and hierarchical RL approaches in both efficiency and success rate. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:35:42 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Hwang",
"Jaebak",
""
],
[
"Lee",
"Sanghyeon",
""
],
[
"Kim",
"Jeongmo",
""
],
[
"Han",
"Seungyul",
""
]
] |
2506.21040 | Shun Inoue | Shun Inoue, Teruaki Enoto, Yuta Notsu, Hiroyuki Uchida, Wataru Buz Iwakiri, Kosuke Namekata, Keith Gendreau | Systematic NICER study of the low-ionized Fe K$\alpha$ line on RS Canum Venaticorum type stars | 19 pages, 12 figures, Accepted for publication in Monthly Notices of the Royal Astronomical Society | null | null | null | astro-ph.SR astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Fe K$\alpha$ fluorescence line ($\sim 6.4$ keV) has been observed during solar and stellar flares. Two emission mechanisms of the Fe K$\alpha$ line, photoionization and collisional ionization, have been discussed, and the aim of this work is to collect evidences for each mechanism employing a statistical correlation approach between the Fe K$\alpha$ line flux and rough flare properties. Here, we systematically searched the NICER (0.2$-$12 keV) archive data for the Fe K$\alpha$ line of RS Canum Venaticorum type stars. Among our analyzed 255 observation IDs with a total exposure of $\sim 700$ ks, we found 25 data sets (total $\sim 40$ ks) exhibiting the Fe K$\alpha$ emission line at 6.37$-$6.54 keV with its equivalent width of 44.3$-$578.4 eV: 18 observations during flares, 6 observations during unconfirmed possible flare candidates and one at a quiescent phase. These observations indicate a positive correlation between the Fe K$\alpha$ line intensity and the 7.11$-$20 keV thermal plasma luminosity with its powerlaw index of $0.86 \pm 0.46$. This correlation in the range of the thermal plasma luminosity $10^{29-33}$ erg s$^{-1}$ is consistent with the photoionization origin of the line. On the other hand, the equivalent width of the Fe K$\alpha$ line has a negative correlation with the 7.11$-$20 keV thermal plasma luminosity with its powerlaw index of $-0.27 \pm 0.10$. This anti-correlation is consistent with the decline of the fluorescence efficiency with increasing the stellar flare loop height. Furthermore, we found a signature of an absorption line at $6.38^{+0.03}_{-0.04}$ keV during a superflare of $\sigma$ Gem. The equivalent width of the line was $-34.7^{+2.03}_{-1.58}$ eV. We discuss the density of the Fe ions from the equivalent width using the curve of growth analysis. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:41:00 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Inoue",
"Shun",
""
],
[
"Enoto",
"Teruaki",
""
],
[
"Notsu",
"Yuta",
""
],
[
"Uchida",
"Hiroyuki",
""
],
[
"Iwakiri",
"Wataru Buz",
""
],
[
"Namekata",
"Kosuke",
""
],
[
"Gendreau",
"Keith",
""
]
] |
2506.21041 | Junwei You | Junwei You, Pei Li, Zhuoyu Jiang, Zilin Huang, Rui Gan, Haotian Shi, Bin Ran | V2X-REALM: Vision-Language Model-Based Robust End-to-End Cooperative Autonomous Driving with Adaptive Long-Tail Modeling | null | null | null | null | cs.RO cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ensuring robust planning and decision-making under rare, diverse, and visually degraded long-tail scenarios remains a fundamental challenge for autonomous driving in urban environments. This issue becomes more critical in cooperative settings, where vehicles and infrastructure jointly perceive and reason across complex environments. To address this challenge, we propose V2X-REALM, a vision-language model (VLM)-based framework with adaptive multimodal learning for robust cooperative autonomous driving under long-tail scenarios. V2X-REALM introduces three core innovations: (i) a prompt-driven long-tail scenario generation and evaluation pipeline that leverages foundation models to synthesize realistic long-tail conditions such as snow and fog across vehicle- and infrastructure-side views, enriching training diversity efficiently; (ii) a gated multi-scenario adaptive attention module that modulates the visual stream using scenario priors to recalibrate ambiguous or corrupted features; and (iii) a multi-task scenario-aware contrastive learning objective that improves multimodal alignment and promotes cross-scenario feature separability. Extensive experiments demonstrate that V2X-REALM significantly outperforms existing baselines in robustness, semantic reasoning, safety, and planning accuracy under complex, challenging driving conditions, advancing the scalability of end-to-end cooperative autonomous driving. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:42:03 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"You",
"Junwei",
""
],
[
"Li",
"Pei",
""
],
[
"Jiang",
"Zhuoyu",
""
],
[
"Huang",
"Zilin",
""
],
[
"Gan",
"Rui",
""
],
[
"Shi",
"Haotian",
""
],
[
"Ran",
"Bin",
""
]
] |
2506.21042 | Boyong He | Boyong He, Yuxiang Ji, Zhuoyue Tan and Liaoni Wu | Boosting Domain Generalized and Adaptive Detection with Diffusion Models: Fitness, Generalization, and Transferability | Accepted by ICCV2025. arXiv admin note: text overlap with arXiv:2503.02101 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Detectors often suffer from performance drop due to domain gap between training and testing data. Recent methods explore diffusion models applied to domain generalization (DG) and adaptation (DA) tasks, but still struggle with large inference costs and have not yet fully leveraged the capabilities of diffusion models. We propose to tackle these problems by extracting intermediate features from a single-step diffusion process, improving feature collection and fusion to reduce inference time by 75% while enhancing performance on source domains (i.e., Fitness). Then, we construct an object-centered auxiliary branch by applying box-masked images with class prompts to extract robust and domain-invariant features that focus on object. We also apply consistency loss to align the auxiliary and ordinary branch, balancing fitness and generalization while preventing overfitting and improving performance on target domains (i.e., Generalization). Furthermore, within a unified framework, standard detectors are guided by diffusion detectors through feature-level and object-level alignment on source domains (for DG) and unlabeled target domains (for DA), thereby improving cross-domain detection performance (i.e., Transferability). Our method achieves competitive results on 3 DA benchmarks and 5 DG benchmarks. Additionally, experiments on COCO generalization benchmark demonstrate that our method maintains significant advantages and show remarkable efficiency in large domain shifts and low-data scenarios. Our work shows the superiority of applying diffusion models to domain generalized and adaptive detection tasks and offers valuable insights for visual perception tasks across diverse domains. The code is available at \href{https://github.com/heboyong/Fitness-Generalization-Transferability}{Fitness-Generalization-Transferability}. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:42:23 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"He",
"Boyong",
""
],
[
"Ji",
"Yuxiang",
""
],
[
"Tan",
"Zhuoyue",
""
],
[
"Wu",
"Liaoni",
""
]
] |
2506.21043 | Shweta Pal | Shweta Pal, Arun Kumar, Monika Agrawal | Analysis of Null Related Beampattern Measures and Signal Quantization Effects for Linear Differential Microphone Arrays | 10 pages, 15 Figures, 3 Tables | null | null | null | eess.SP | http://creativecommons.org/licenses/by-nc-nd/4.0/ | A differential microphone array (DMA) offers enhanced capabilities to obtain sharp nulls at the cost of relatively broad peaks in the beam power pattern. This can be used for applications that require nullification or attenuation of interfering sources. To the best of our knowledge, the existing literature lacks measures that directly assess the efficacy of nulls, and null-related measures have not been investigated in the context of differential microphone arrays (DMAs). This paper offers new insights about the utility of DMAs by proposing measures that characterize the nulls in their beam power patterns. We investigate the performance of differential beamformers by presenting and evaluating null-related measures namely null depth (ND) and Null Width (NW) as a function of depth level relative to the beam power pattern maxima. A study of signal quantization effects due to data acquisition for 1st, 2nd and 3rd order linear DMAs and for different beampatterns i.e. dipole, cardioid, hypercardioid and supercardioid is presented. An analytical expression for the quantized beamformed output for any general $ N^{th} $ order DMA is formulated. Simulation results of the variation of ND with number of quantization bits and the variation of NW as a function of depth are also presented and inferences are drawn. Lab experiments are conducted in a fully anechoic room to support the simulation results. The measured beampattern exhibits a pronounced null depth, confirming the effectiveness of the experimental setup. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:45:25 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Pal",
"Shweta",
""
],
[
"Kumar",
"Arun",
""
],
[
"Agrawal",
"Monika",
""
]
] |
2506.21044 | He Zhang | He Zhang, Ming Zhou, Shaopeng Zhai, Ying Sun, Hui Xiong | Efficient Skill Discovery via Regret-Aware Optimization | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Unsupervised skill discovery aims to learn diverse and distinguishable behaviors in open-ended reinforcement learning. For existing methods, they focus on improving diversity through pure exploration, mutual information optimization, and learning temporal representation. Despite that they perform well on exploration, they remain limited in terms of efficiency, especially for the high-dimensional situations. In this work, we frame skill discovery as a min-max game of skill generation and policy learning, proposing a regret-aware method on top of temporal representation learning that expands the discovered skill space along the direction of upgradable policy strength. The key insight behind the proposed method is that the skill discovery is adversarial to the policy learning, i.e., skills with weak strength should be further explored while less exploration for the skills with converged strength. As an implementation, we score the degree of strength convergence with regret, and guide the skill discovery with a learnable skill generator. To avoid degeneration, skill generation comes from an up-gradable population of skill generators. We conduct experiments on environments with varying complexities and dimension sizes. Empirical results show that our method outperforms baselines in both efficiency and diversity. Moreover, our method achieves a 15% zero shot improvement in high-dimensional environments, compared to existing methods. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:45:59 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Zhang",
"He",
""
],
[
"Zhou",
"Ming",
""
],
[
"Zhai",
"Shaopeng",
""
],
[
"Sun",
"Ying",
""
],
[
"Xiong",
"Hui",
""
]
] |
2506.21045 | Hansam Cho | Hansam Cho, Seoung Bum Kim | Improving Diffusion-Based Image Editing Faithfulness via Guidance and Scheduling | preprint | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Text-guided diffusion models have become essential for high-quality image synthesis, enabling dynamic image editing. In image editing, two crucial aspects are editability, which determines the extent of modification, and faithfulness, which reflects how well unaltered elements are preserved. However, achieving optimal results is challenging because of the inherent trade-off between editability and faithfulness. To address this, we propose Faithfulness Guidance and Scheduling (FGS), which enhances faithfulness with minimal impact on editability. FGS incorporates faithfulness guidance to strengthen the preservation of input image information and introduces a scheduling strategy to resolve misalignment between editability and faithfulness. Experimental results demonstrate that FGS achieves superior faithfulness while maintaining editability. Moreover, its compatibility with various editing methods enables precise, high-quality image edits across diverse tasks. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:46:03 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Cho",
"Hansam",
""
],
[
"Kim",
"Seoung Bum",
""
]
] |
2506.21046 | Shangbo Wu | Shangbo Wu, Yu-an Tan, Ruinan Ma, Wencong Ma, Dehua Zhu, Yuanzhang Li | Boosting Generative Adversarial Transferability with Self-supervised Vision Transformer Features | 14 pages, 9 figures, to appear in ICCV 2025 | null | null | null | cs.CV cs.CR | http://creativecommons.org/licenses/by-sa/4.0/ | The ability of deep neural networks (DNNs) come from extracting and interpreting features from the data provided. By exploiting intermediate features in DNNs instead of relying on hard labels, we craft adversarial perturbation that generalize more effectively, boosting black-box transferability. These features ubiquitously come from supervised learning in previous work. Inspired by the exceptional synergy between self-supervised learning and the Transformer architecture, this paper explores whether exploiting self-supervised Vision Transformer (ViT) representations can improve adversarial transferability. We present dSVA -- a generative dual self-supervised ViT features attack, that exploits both global structural features from contrastive learning (CL) and local textural features from masked image modeling (MIM), the self-supervised learning paradigm duo for ViTs. We design a novel generative training framework that incorporates a generator to create black-box adversarial examples, and strategies to train the generator by exploiting joint features and the attention mechanism of self-supervised ViTs. Our findings show that CL and MIM enable ViTs to attend to distinct feature tendencies, which, when exploited in tandem, boast great adversarial generalizability. By disrupting dual deep features distilled by self-supervised ViTs, we are rewarded with remarkable black-box transferability to models of various architectures that outperform state-of-the-arts. Code available at https://github.com/spencerwooo/dSVA. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:47:51 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Wu",
"Shangbo",
""
],
[
"Tan",
"Yu-an",
""
],
[
"Ma",
"Ruinan",
""
],
[
"Ma",
"Wencong",
""
],
[
"Zhu",
"Dehua",
""
],
[
"Li",
"Yuanzhang",
""
]
] |
2506.21047 | Dao Xiang | Tianyu Wang, Hui Jiang, Ming Zhang, Xiao Zou, Pengfei Zhu, Feng He, Zheng Li, and Dao Xiang | Probing valence electron and hydrogen dynamics using charge-pair imaging with ultrafast electron diffraction | null | null | null | null | physics.chem-ph physics.atm-clus physics.atom-ph physics.optics | http://creativecommons.org/licenses/by/4.0/ | A key challenge in ultrafast science has been to directly track the coupled motions of electrons and nuclei in real-space and real-time. This study presents a significant step towards this goal by demonstrating the feasibility of time-resolved real-space tracking of valence electron and hydrogen dynamics during the photodissociation of ammonia (NH3) using MeV ultrafast electron diffraction. It is demonstrated that the enhanced temporal resolution, in conjunction with the analysis of the charge-pair distribution function, enables the disentanglement of the correlated motion of valence electrons and hydrogens in photoexcited ammonia molecule. The methodology employed in this study, which utilizes the charge-pair distribution function from ultrafast electron scattering to retrieve intertwined electron and nucleus dynamics, may open up new opportunities in the study of quantum dynamics for a wide range of molecules. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:48:16 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Wang",
"Tianyu",
""
],
[
"Jiang",
"Hui",
""
],
[
"Zhang",
"Ming",
""
],
[
"Zou",
"Xiao",
""
],
[
"Zhu",
"Pengfei",
""
],
[
"He",
"Feng",
""
],
[
"Li",
"Zheng",
""
],
[
"Xiang",
"Dao",
""
]
] |
2506.21048 | Salvatore Capozziello | Salvatore Capozziello, Giuseppe Meluccio, and Jonas R. Mureika | Uncertainty Principles and Non-local Black Holes | 5 pages | null | null | null | gr-qc hep-th | http://creativecommons.org/licenses/by/4.0/ | We discuss the Generalized Uncertainty Principle and the Extended Uncertainty Principle in the context of black hole solutions coming from non-local theories of gravity, focusing, specifically, on Infinite Derivative Gravity. We argue that these modifications of the Heisenberg Uncertainty Principle are effective descriptions arising from the non-local features of gravitational interaction. By comparing the predictions of both the modified uncertainty principles and non-local gravity, we find theoretical constraints on otherwise free parameters as well as universal laws for black hole physics beyond General Relativity. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:51:14 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Capozziello",
"Salvatore",
""
],
[
"Meluccio",
"Giuseppe",
""
],
[
"Mureika",
"Jonas R.",
""
]
] |
2506.21049 | Chun Yuan Yuan | Chunyuan Yuan, Chong Zhang, Zheng Fang, Ming Pang, Xue Jiang, Changping Peng, Zhangang Lin, Ching Law | A Semi-supervised Scalable Unified Framework for E-commerce Query Classification | Accepted by ACL 2025 | null | null | null | cs.CL cs.AI cs.IR | http://creativecommons.org/licenses/by/4.0/ | Query classification, including multiple subtasks such as intent and category prediction, is vital to e-commerce applications. E-commerce queries are usually short and lack context, and the information between labels cannot be used, resulting in insufficient prior information for modeling. Most existing industrial query classification methods rely on users' posterior click behavior to construct training samples, resulting in a Matthew vicious cycle. Furthermore, the subtasks of query classification lack a unified framework, leading to low efficiency for algorithm optimization.
In this paper, we propose a novel Semi-supervised Scalable Unified Framework (SSUF), containing multiple enhanced modules to unify the query classification tasks. The knowledge-enhanced module uses world knowledge to enhance query representations and solve the problem of insufficient query information. The label-enhanced module uses label semantics and semi-supervised signals to reduce the dependence on posterior labels. The structure-enhanced module enhances the label representation based on the complex label relations. Each module is highly pluggable, and input features can be added or removed as needed according to each subtask. We conduct extensive offline and online A/B experiments, and the results show that SSUF significantly outperforms the state-of-the-art models. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:52:33 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yuan",
"Chunyuan",
""
],
[
"Zhang",
"Chong",
""
],
[
"Fang",
"Zheng",
""
],
[
"Pang",
"Ming",
""
],
[
"Jiang",
"Xue",
""
],
[
"Peng",
"Changping",
""
],
[
"Lin",
"Zhangang",
""
],
[
"Law",
"Ching",
""
]
] |
2506.21050 | Ryuichi Shindou | Pengwei Zhao, and Ryuichi Shindou | Theory of the Anderson transition in three-dimensional chiral symmetry classes: Connection to type-II superconductors | 18 pages, 2 figures | null | null | null | cond-mat.dis-nn cond-mat.stat-mech cond-mat.supr-con | http://creativecommons.org/licenses/by/4.0/ | Phase transitions governed by topological defects constitute a cornerstone of modern physics. Two-dimensional (2D) Anderson transitions in chiral symmetry classes are driven by the proliferation of vortex-antivortex pairs -- a mechanism analogous to the Berezinskii-Kosterlitz-Thouless (BKT) transition in the 2D XY model. In this work, we extend this paradigm to three-dimensional (3D) chiral symmetry classes, where vortex loops emerge as the key topological defects governing the Anderson transition. By deriving the dual representation of the 3D nonlinear sigma model for the chiral unitary class, we develop a mean-field theory of its Anderson transition and elucidate the role of 1D weak band topology in the Anderson transition. Strikingly, our dual representation of the 3D NLSM in the chiral symmetry class uncovers its connection to the magnetostatics of 3D type-II superconductors. The metal-to-quasilocalized and quasilocalized-to-insulating transitions in 3D chiral symmetry class share a unified theoretical framework with the normal-to-mixed and mixed-to-superconducting transitions in 3D type-II superconductors under an external magnetic field, respectively. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:52:47 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Zhao",
"Pengwei",
""
],
[
"Shindou",
"Ryuichi",
""
]
] |
2506.21051 | Mingxing Luo | Yan-Han Yang, Xin-Zhu Liu, Xing-Zhou Zheng, Jun-Li Jiang, Xue Yang, Shao-Ming Fei, Zhihao Ma, Zizhu Wang, Ming-Xing Luo | Experimental Characterization of Quantumness Using the Uncertainty Principle, Coherence, and Nonlocality | 5+7 pages, 7 figures | Physical Review Letters, 2025 | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Heisenberg's uncertainty principle, coherence and Bell nonlocality have been individually examined through many experiments. In this Letter, we systematically characterize all of this quantumness in a unified manner. We first construct universal uncertainty relations to reveal intrinsic features of incompatible measurements, which include all the state-independent uncertainties as special cases. We further extend to witness both quantum coherence and Bell nonlocality. We finally perform experiments with unified two-photon states, and validate the uncertainty principle, coherence and Bell nonlocality within the experimental error. Our methods for witnessing quantumness are valuable in characterizing quantum correlations in quantum information processing. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:53:40 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yang",
"Yan-Han",
""
],
[
"Liu",
"Xin-Zhu",
""
],
[
"Zheng",
"Xing-Zhou",
""
],
[
"Jiang",
"Jun-Li",
""
],
[
"Yang",
"Xue",
""
],
[
"Fei",
"Shao-Ming",
""
],
[
"Ma",
"Zhihao",
""
],
[
"Wang",
"Zizhu",
""
],
[
"Luo",
"Ming-Xing",
""
]
] |
2506.21052 | Hugh Dennin | Hugh Dennin | Cauchy identities for Grothendieck polynomials and a dual RSK correspondence through pipe dreams | 36 pages, 19 figures | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | The Cauchy identity gives a recipe for decomposing a double Grothendieck polynomial $\mathfrak{G}^{(\beta)}_w(x;y)$ as a sum of products $\mathfrak{G}^{(\beta)}_v(x)\mathfrak{G}^{(\beta)}_u(y)$ of single Grothendieck polynomials. Combinatorially, this identity suggests the existence of a weight-preserving bijection between certain families of diagrams called pipe dreams. In this paper, we provide such a bijection using an algorithm called pipe dream rectification. In turn, rectification is built from a new class of flow operators which themselves exhibit a surprising symmetry. Finally, we examine other applications of rectification including an insertion algorithm on pipe dreams which recovers a variant of the dual RSK correspondence. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:58:47 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Dennin",
"Hugh",
""
]
] |
2506.21053 | Fuqiang Niu | Fuqiang Niu, Genan Dai, Yisha Lu, Jiayu Liao, Xiang Li, Hu Huang and Bowen Zhang | MT2-CSD: A New Dataset and Multi-Semantic Knowledge Fusion Method for Conversational Stance Detection | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the realm of contemporary social media, automatic stance detection is pivotal for opinion mining, as it synthesizes and examines user perspectives on contentious topics to uncover prevailing trends and sentiments. Traditional stance detection research often targets individual instances, thereby limiting its capacity to model multi-party discussions typical in real social media scenarios. This shortcoming largely stems from the scarcity of datasets that authentically capture the dynamics of social media interactions, hindering advancements in conversational stance detection. In this paper, we introduce MT2-CSD, a comprehensive dataset for multi-target, multi-turn conversational stance detection. To the best of our knowledge, MT2-CSD is the largest dataset available for this purpose, comprising 24,457 annotated instances and exhibiting the greatest conversational depth, thereby presenting new challenges for stance detection. To address these challenges, we propose the Large Language model enhanced Conversational Relational Attention Network (LLM-CRAN), which exploits the reasoning capabilities of LLMs to improve conversational understanding. We conduct extensive experiments to evaluate the efficacy of LLM-CRAN on the MT2-CSD dataset. The experimental results indicate that LLM-CRAN significantly outperforms strong baseline models in the task of conversational stance detection. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 06:59:30 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Niu",
"Fuqiang",
""
],
[
"Dai",
"Genan",
""
],
[
"Lu",
"Yisha",
""
],
[
"Liao",
"Jiayu",
""
],
[
"Li",
"Xiang",
""
],
[
"Huang",
"Hu",
""
],
[
"Zhang",
"Bowen",
""
]
] |
2506.21054 | Fu Peng | Fu Peng, Ming Tang | FedDAA: Dynamic Client Clustering for Concept Drift Adaptation in Federated Learning | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | In federated learning (FL), the data distribution of each client may change over time, introducing both temporal and spatial data heterogeneity, known as concept drift. Data heterogeneity arises from three drift sources: real drift (a shift in the conditional distribution P(y|x)), virtual drift (a shift in the input distribution P(x)), and label drift (a shift in the label distribution P(y)). However, most existing FL methods addressing concept drift primarily focus on real drift. When clients experience virtual or label drift, these methods often fail to selectively retain useful historical knowledge, leading to catastrophic forgetting. A key challenge lies in distinguishing different sources of drift, as they require distinct adaptation strategies: real drift calls for discarding outdated data, while virtual or label drift benefits from retaining historical data. Without explicitly identifying the drift sources, a general adaptation strategy is suboptimal and may harm generalization. To address this challenge, we propose FedDAA, a dynamic clustered FL framework designed to adapt to multi-source concept drift while preserving valuable historical knowledge. Specifically, FedDAA integrates three modules: a cluster number determination module to find the optimal number of clusters; a real drift detection module to distinguish real drift from virtual/label drift; and a concept drift adaptation module to adapt to new data while retaining useful historical information. We provide theoretical convergence guarantees, and experiments show that FedDAA achieves 7.84% to 8.52% accuracy improvements over state-of-the-art methods on Fashion-MNIST, CIFAR-10, and CIFAR-100. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:09:08 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Peng",
"Fu",
""
],
[
"Tang",
"Ming",
""
]
] |
2506.21055 | Jiahao Lyu | Demin Zhang, Jiahao Lyu, Zhijie Shen, Yu Zhou | Class-Agnostic Region-of-Interest Matching in Document Images | Accepted by ICDAR2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Document understanding and analysis have received a lot of attention due to their widespread application. However, existing document analysis solutions, such as document layout analysis and key information extraction, are only suitable for fixed category definitions and granularities, and cannot achieve flexible applications customized by users. Therefore, this paper defines a new task named ``Class-Agnostic Region-of-Interest Matching'' (``RoI-Matching'' for short), which aims to match the customized regions in a flexible, efficient, multi-granularity, and open-set manner. The visual prompt of the reference document and target document images are fed into our model, while the output is the corresponding bounding boxes in the target document images. To meet the above requirements, we construct a benchmark RoI-Matching-Bench, which sets three levels of difficulties following real-world conditions, and propose the macro and micro metrics to evaluate. Furthermore, we also propose a new framework RoI-Matcher, which employs a siamese network to extract multi-level features both in the reference and target domains, and cross-attention layers to integrate and align similar semantics in different domains. Experiments show that our method with a simple procedure is effective on RoI-Matching-Bench, and serves as the baseline for further research. The code is available at https://github.com/pd162/RoI-Matching. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:09:19 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Zhang",
"Demin",
""
],
[
"Lyu",
"Jiahao",
""
],
[
"Shen",
"Zhijie",
""
],
[
"Zhou",
"Yu",
""
]
] |
2506.21056 | Trung Nghia Le | Dinh-Khoi Vo and Van-Loc Nguyen and Minh-Triet Tran and Trung-Nghia Le | SAMURAI: Shape-Aware Multimodal Retrieval for 3D Object Identification | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Retrieving 3D objects in complex indoor environments using only a masked 2D image and a natural language description presents significant challenges. The ROOMELSA challenge limits access to full 3D scene context, complicating reasoning about object appearance, geometry, and semantics. These challenges are intensified by distorted viewpoints, textureless masked regions, ambiguous language prompts, and noisy segmentation masks. To address this, we propose SAMURAI: Shape-Aware Multimodal Retrieval for 3D Object Identification. SAMURAI integrates CLIP-based semantic matching with shape-guided re-ranking derived from binary silhouettes of masked regions, alongside a robust majority voting strategy. A dedicated preprocessing pipeline enhances mask quality by extracting the largest connected component and removing background noise. Our hybrid retrieval framework leverages both language and shape cues, achieving competitive performance on the ROOMELSA private test set. These results highlight the importance of combining shape priors with language understanding for robust open-world 3D object retrieval. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:10:52 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Vo",
"Dinh-Khoi",
""
],
[
"Nguyen",
"Van-Loc",
""
],
[
"Tran",
"Minh-Triet",
""
],
[
"Le",
"Trung-Nghia",
""
]
] |
2506.21057 | Zhuochen Miao | Zhuochen Miao, Jun Lv, Hongjie Fang, Yang Jin, Cewu Lu | Knowledge-Driven Imitation Learning: Enabling Generalization Across Diverse Conditions | IROS 2025 | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Imitation learning has emerged as a powerful paradigm in robot manipulation, yet its generalization capability remains constrained by object-specific dependencies in limited expert demonstrations. To address this challenge, we propose knowledge-driven imitation learning, a framework that leverages external structural semantic knowledge to abstract object representations within the same category. We introduce a novel semantic keypoint graph as a knowledge template and develop a coarse-to-fine template-matching algorithm that optimizes both structural consistency and semantic similarity. Evaluated on three real-world robotic manipulation tasks, our method achieves superior performance, surpassing image-based diffusion policies with only one-quarter of the expert demonstrations. Extensive experiments further demonstrate its robustness across novel objects, backgrounds, and lighting conditions. This work pioneers a knowledge-driven approach to data-efficient robotic learning in real-world settings. Code and more materials are available on https://knowledge-driven.github.io/. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:14:59 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Miao",
"Zhuochen",
""
],
[
"Lv",
"Jun",
""
],
[
"Fang",
"Hongjie",
""
],
[
"Jin",
"Yang",
""
],
[
"Lu",
"Cewu",
""
]
] |
2506.21058 | Yan V. Fyodorov | Yan V Fyodorov | Kac-Rice inspired approach to non-Hermitian random matrices | 41 page, 1 figure | null | null | null | math-ph cond-mat.dis-nn math.MP math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We suggest a method of analyzing the joint probability density (JPD) ${\cal P}_N(z,{\bf v})$ of an eigenvalue $z$ and the associated right eigenvector ${\bf v}$ (normalized with ${\bf v}^*{\bf v}=1$) of non-Hermitian random matrices of a given size $N\times N$. The method, which represents an alternative to Girko Hermitization approach, is essentially based on the Kac-Rice counting formula applied to the associated characteristic polynomial combined with a certain integral identity for the Dirac delta function of such a polynomial. To illustrate utility of the general method we apply it to consider two particular cases: (i) one-parameter family of matrices interpolating between complex Ginibre and real Ginibre ensembles and (ii) a complex Ginibre matrix additively perturbed by a fixed matrix. In particular, in the former case we discuss the formation of an excess of eigenvalues in the vicinity of the real axis on approaching the real Ginibre limit, which eventually gives rise to the existence of a new scaling regime of "weak non-reality" as $N\to \infty$. In the second case we provide new insights into eigenvalue and eigenvector distribution for a general rank one perturbation of complex Ginibre matrices of finite size $N$, and in the structure of an outlier as $N\gg 1$. Finally we discuss a possible generalization of the proposed method which is expected to be suitable for analysis of JPD involving both left- and right eigenvectors. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:16:22 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Fyodorov",
"Yan V",
""
]
] |
2506.21059 | David Valderrama | David Valderrama | Splitting Families, Reaping Families, and Families of Permutations Associated with Asymptotic Density | null | null | null | null | math.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate several relations between cardinal characteristics of the continuum related with the asymptotic density of the natural numbers and some known cardinal invariants. Specifically, we study the cardinals of the form $\mathfrak{s}_X$, $\mathfrak{r}_X$ and $\mathfrak{dd}_{X,Y}$ introduced in arXiv:2304.09698 and arXiv:2410.21102, answering some questions raised in these papers. In particular, we prove that $\mathfrak{s}_0=$ cov$(\mathcal{M})$ and $\mathfrak{r}_0=$ non$(\mathcal{M})$. We also show that $\mathfrak{dd}_{\{r\}, \textsf{all}}=\mathfrak{dd}_{\{1/2\}, \textsf{all}}$ for all $r\in (0,1)$, and we provide a proof of Con($\mathfrak{dd}_{(0,1),\{0,1\}}^{\textsf{rel}}<$ non$(\mathcal{N})$) and Con($\mathfrak{dd}_{\textsf{all},\textsf{all}}^{\textsf{rel}}<$ non$(\mathcal{N})$). | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:16:33 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Valderrama",
"David",
""
]
] |
2506.21060 | Mingxing Luo | Jun-Li Jiang, Xin-Zhu Liu, Xue Yang, Xiuyong Ding, Da Zhang, Ming-Xing Luo | One-way network nonlocality of continuous variable entangled networks | 8 pages, 4 figures | Advanced Quantum Techniques, 2025 | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Nonlocality is a key feature of quantum networks and is being studied for its potential applications in quantum communication and computing. Understanding and harnessing nonlocality in quantum networks could lead to the development of faster and more secure communication systems. All the nonclassicalities are limited to discrete variable quantum networks. We propose the first method to verify the network nonlocality of all optical quantum network consisting of two entangled states where one-way classical communication is allowed. This provides the first device-independent method to verify the quantum correlations generated from all optical continuous-variable quantum networks. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:18:15 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Jiang",
"Jun-Li",
""
],
[
"Liu",
"Xin-Zhu",
""
],
[
"Yang",
"Xue",
""
],
[
"Ding",
"Xiuyong",
""
],
[
"Zhang",
"Da",
""
],
[
"Luo",
"Ming-Xing",
""
]
] |
2506.21061 | Zi-Yong Ge | Zhiguang Yan, Zi-Yong Ge, Rui Li, Yu-Ran Zhang, Franco Nori, and Yasunobu Nakamura | Characterizing Many-body Dynamics with Projected Ensembles on a Superconducting Quantum Processor | 24+31 pages, 4 figures | null | null | null | quant-ph cond-mat.stat-mech | http://creativecommons.org/licenses/by/4.0/ | Quantum simulators offer a new opportunity for the experimental exploration of non-equilibrium quantum many-body dynamics, which have traditionally been characterized through expectation values or entanglement measures, based on density matrices of the system. Recently, a more general framework for studying quantum many-body systems based on projected ensembles has been introduced, revealing novel quantum phenomena, such as deep thermalization in chaotic systems. Here, we experimentally investigate a chaotic quantum many-body system using projected ensembles on a superconducting processor with 16 qubits on a square lattice. Our results provide direct evidence of deep thermalization by observing a Haar-distributed projected ensemble for the steady states within a charge-conserved sector. Moreover, by introducing an ensemble-averaged entropy as a metric, we establish a benchmark for many-body information leakage from the system to its environment. Our work paves the way for studying quantum many-body dynamics using projected ensembles and shows a potential implication for advancing quantum simulation techniques. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:19:48 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yan",
"Zhiguang",
""
],
[
"Ge",
"Zi-Yong",
""
],
[
"Li",
"Rui",
""
],
[
"Zhang",
"Yu-Ran",
""
],
[
"Nori",
"Franco",
""
],
[
"Nakamura",
"Yasunobu",
""
]
] |
2506.21062 | Tomoo Yokoyama | Tomoo Yokoyama | Structural stability and generic transitions of "incompressible" line fields on surfaces | null | null | null | null | math.DS math.GT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Various line fields naturally arise on surfaces in both physical and biological contexts, and generic singularities frequently appear in the form of 1-prong (thorn-like) and 3-prong (tripod-like) configurations, which can be modeled by partial differential equations with specific parameter values. However, it remains open under which topologies such line fields are structurally stable and form an open dense subset. In this paper, we propose a new topological framework for describing line fields and their evaluations on surfaces that is suitable from both theoretical and applied perspectives. Specifically, we demonstrate that, under a topology defined by a ``cone'' structure, line fields with 1-prong and 3-prong singularities are generic when an ``incompressibility condition'' holds. We also introduce representations of complete invariants for generic line fields and their generic transitions. These representations enable the evolution of ``incompressible'' line fields -- such as those observed in active nematics -- to be encoded as walks on transition graphs, providing a combinatorial framework for their analysis. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:25:05 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yokoyama",
"Tomoo",
""
]
] |
2506.21063 | Lin Hong | Lin Hong, Lu Liu, Zhouhua Peng, and Fumin Zhang | Control of Marine Robots in the Era of Data-Driven Intelligence | null | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | The control of marine robots has long relied on model-based methods grounded in classical and modern control theory. However, the nonlinearity and uncertainties inherent in robot dynamics, coupled with the complexity of marine environments, have revealed the limitations of conventional control methods. The rapid evolution of machine learning has opened new avenues for incorporating data-driven intelligence into control strategies, prompting a paradigm shift in the control of marine robots. This paper provides a review of recent progress in marine robot control through the lens of this emerging paradigm. The review covers both individual and cooperative marine robotic systems, highlighting notable achievements in data-driven control of marine robots and summarizing open-source resources that support the development and validation of advanced control methods. Finally, several future perspectives are outlined to guide research toward achieving high-level autonomy for marine robots in real-world applications. This paper aims to serve as a roadmap toward the next-generation control framework of marine robots in the era of data-driven intelligence. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:28:03 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Hong",
"Lin",
""
],
[
"Liu",
"Lu",
""
],
[
"Peng",
"Zhouhua",
""
],
[
"Zhang",
"Fumin",
""
]
] |
2506.21064 | Brendan Pawlowski | Sara C. Billey, Yibo Gao, Brendan Pawlowski | Introduction to the Cohomology of the Flag Variety | 192 pages | null | null | null | math.CO math.AC math.AG | http://creativecommons.org/licenses/by/4.0/ | One hundred years ago, Hilbert gave a list of important open problems in mathematics. His 15th problem asked for the development of a rigorous calculus explaining Schubert's enumerative results for intersecting varieties defined by rank conditions on vector spaces. Today by way of many contributions in algebraic topology, geometry, and combinatorics, we consider this solved. Yet, deep questions remain about the subtleties of actually carrying out the process. In this chapter, we hope to summarize the rigorous development of what has become known as Schubert calculus, with an eye toward computation. We discuss Grassmannians and flag varieties and their cohomology rings, following Monk's constructive algebraic approach. We derive formulas for Schur and Schubert polynomials, which represent cohomology classes of Schubert varieties. We hint at the vast literature in this area and point to the other references in the Handbook for more information. Finally, we identify open problems that remain a challenge even with modern tools at our fingertips in hopes of inspiring further contributions in this fascinating field.
This is intended as the first chapter of a book entitled "Handbook of Combinatorial Algebraic Geometry: Subvarieties of the Flag Variety", a compendium of topics in the area. The book is being edited by Erik Insko, Martha Precup, and Ed Richmond. In addition to this introductory chapter, others will cover more advanced topics such as Kazhdan-Lusztig varieties, generalized smooth Schubert varieties, Richardson varieties and positroid varieties, spherical and torus orbit closures, spanning line configurations, different types of Hessenberg varieties, and generalizations to Kac-Moody flag varieties, each written by experts in those areas. We hope you enjoy this chapter enough to seek out the others, and that you send us any comments or corrections you find as you read this article! | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:33:38 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Billey",
"Sara C.",
""
],
[
"Gao",
"Yibo",
""
],
[
"Pawlowski",
"Brendan",
""
]
] |
2506.21065 | Magnus Sv\"ard | Magnus Sv\"ard, Anita Gjesteland | Entropy-stable in- and outflow boundary conditions for the compressible Navier-Stokes equations | null | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose inflow and outflow boundary conditions for the compressible Navier-Stokes equations and prove that they allow a priori estimates of the entropy, mass and total energy. Furthermore, we demonstrate how to approximate these boundary conditions in conjunction with an entropy-stable finite-volume scheme. The method is also applicable to other types of entropy-stable schemes. Finally, we carry out some numerical computations with the finite-volume scheme and demonstrate their robustness. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:34:21 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Svärd",
"Magnus",
""
],
[
"Gjesteland",
"Anita",
""
]
] |
2506.21066 | Tarak Patra | Rahul Karmakar, Abhishek S. Chankapure, Srikanth Sastry, Sanat K. Kumar and Tarak K. Patra | Microphase Separation Controls the Dynamics of Associative Vitrimers | null | null | null | null | cond-mat.soft | http://creativecommons.org/licenses/by/4.0/ | Vitrimers are a class of polymers characterized by dynamic covalent networks, where specific monomer units, which are know as stickers, form reversible crosslinks that enable network rearrangement without loss of overall connectivity. The conventional wisdom is that the sticker dynamics control chain relaxation behavior, and hence the mechanical properties of associative vitrimers where the number of crosslinked sticker pairs is precisely constant over time. Instead, here we show that the chemical differences between sticker groups and nonsticky chain monomers can cause them to microphase separate. Under these conditions, the slow exchange of a sticker from one microphase to an adjacent one controls relaxation behavior. Controlling sticker aggregation is thus a key to tailoring the properties of these polymers with immediate relevance to a circular polymer economy. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:35:44 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Karmakar",
"Rahul",
""
],
[
"Chankapure",
"Abhishek S.",
""
],
[
"Sastry",
"Srikanth",
""
],
[
"Kumar",
"Sanat K.",
""
],
[
"Patra",
"Tarak K.",
""
]
] |
2506.21067 | Vahid Asadi | Vahid Asadi, Akram Hasani Zonoozi, Hosein Haghi, Fatemeh Abedini, Atousa Kalantari, Marziye Jafariyazani, Nima Chartab | Leveraging Machine Learning for Accurate and Fast Stellar Mass Estimation of Galaxies | null | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Unveiling the evolutionary history of galaxies necessitates a precise understanding of their physical properties. Traditionally, astronomers achieve this through spectral energy distribution (SED) fitting. However, this approach can be computationally intensive and time-consuming, particularly for large datasets. This study investigates the viability of machine learning (ML) algorithms as an alternative to traditional SED-fitting for estimating stellar masses in galaxies. We compare a diverse range of unsupervised and supervised learning approaches including prominent algorithms such as K-means, HDBSCAN, Parametric t-Distributed Stochastic Neighbor Embedding (Pt-SNE), Principal Component Analysis (PCA), Random Forest, and Self-Organizing Maps (SOM) against the well-established LePhare code, which performs SED-fitting as a benchmark. We train various ML algorithms using simple model SEDs in photometric space, generated with the BC03 code. These trained algorithms are then employed to estimate the stellar masses of galaxies within a subset of the COSMOS survey dataset. The performance of these ML methods is subsequently evaluated and compared with the results obtained from LePhare, focusing on both accuracy and execution time. Our evaluation reveals that ML algorithms can achieve comparable accuracy to LePhare while offering significant speed advantages (1,000 to 100,000 times faster). K-means and HDBSCAN emerge as top performers among our selected ML algorithms. Supervised learning algorithms like Random Forest and manifold learning techniques such as Pt-SNE and SOM also show promising results. These findings suggest that ML algorithms hold significant promise as a viable alternative to traditional SED-fitting methods for estimating the stellar masses of galaxies. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:36:36 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Asadi",
"Vahid",
""
],
[
"Zonoozi",
"Akram Hasani",
""
],
[
"Haghi",
"Hosein",
""
],
[
"Abedini",
"Fatemeh",
""
],
[
"Kalantari",
"Atousa",
""
],
[
"Jafariyazani",
"Marziye",
""
],
[
"Chartab",
"Nima",
""
]
] |
2506.21068 | Ji\v{r}\'i Pavelec | Moritz Eder, Faith J. Lewis, Johanna I. H\"utner, Panukorn Sombut, Maosheng Hao, David Rath, Jan Balajka, Margareta Wagner, Matthias Meier, Cesare Franchini, Ulrike Diebold, Michael Schmid, Florian Libisch, Ji\v{r}\'i Pavelec, and Gareth S. Parkinson | Multi-Technique Characterization of Rhodium Gem-Dicarbonyls on TiO$_2$(110) | null | null | null | null | physics.chem-ph cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Gem-dicarbonyls of transition metals supported on metal (oxide) surfaces are common intermediates in heterogeneous catalysis. While infrared (IR) spectroscopy is a standard tool for detecting these species on applied catalysts, the ill-defined crystallographic environment of species observed on powder catalysts renders data interpretation challenging. In this work, we apply a multi-technique surface science approach to investigate rhodium gem-dicarbonyls on a single-crystalline rutile TiO$_2$(110) surface. We combine spectroscopy, scanning probe microscopy, and Density Functional Theory (DFT) to determine their location and coordination on the surface. IR spectroscopy shows the successful creation of gem-dicarbonyls on a titania single crystal by exposing deposited Rh atoms to CO gas, followed by annealing to 200-250 K. Low-temperature scanning tunneling microscopy (STM) and non-contact atomic force microscopy (nc-AFM) data reveal that these complexes are mostly aligned along the [001] crystallographic direction, corroborating theoretical predictions. Notably, x-ray photoelectron spectroscopy (XPS) data reveal multiple rhodium species on the surface, even when the IR spectra show only the signature of rhodium gem-dicarbonyls. As such, our results highlight the complex behavior of carbonyls on metal oxide surfaces, and demonstrate the necessity of multi-technique approaches for the adequate characterization of single-atom catalysts. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:38:19 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Eder",
"Moritz",
""
],
[
"Lewis",
"Faith J.",
""
],
[
"Hütner",
"Johanna I.",
""
],
[
"Sombut",
"Panukorn",
""
],
[
"Hao",
"Maosheng",
""
],
[
"Rath",
"David",
""
],
[
"Balajka",
"Jan",
""
],
[
"Wagner",
"Margareta",
""
],
[
"Meier",
"Matthias",
""
],
[
"Franchini",
"Cesare",
""
],
[
"Diebold",
"Ulrike",
""
],
[
"Schmid",
"Michael",
""
],
[
"Libisch",
"Florian",
""
],
[
"Pavelec",
"Jiří",
""
],
[
"Parkinson",
"Gareth S.",
""
]
] |
2506.21069 | Xieyang Sun | Xieyang Sun, Yuanqing Zheng, Wei Xi, Zuhao Chen, Zhizhen Chen, Han Hao, Zhiping Jiang, Sheng Zhong | TEMPEST-LoRa: Cross-Technology Covert Communication | 15 pages, 19 figures, and this paper has been accepted to ACM CCS 2025 | null | 10.1145/3719027.3744817 | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Electromagnetic (EM) covert channels pose significant threats to computer and communications security in air-gapped networks. Previous works exploit EM radiation from various components (e.g., video cables, memory buses, CPUs) to secretly send sensitive information. These approaches typically require the attacker to deploy highly specialized receivers near the victim, which limits their real-world impact. This paper reports a new EM covert channel, TEMPEST-LoRa, that builds on Cross-Technology Covert Communication (CTCC), which could allow attackers to covertly transmit EM-modulated secret data from air-gapped networks to widely deployed operational LoRa receivers from afar. We reveal the potential risk and demonstrate the feasibility of CTCC by tackling practical challenges involved in manipulating video cables to precisely generate the EM leakage that could readily be received by third-party commercial LoRa nodes/gateways. Experiment results show that attackers can reliably decode secret data modulated by the EM leakage from a video cable at a maximum distance of 87.5m or a rate of 21.6 kbps. We note that the secret data transmission can be performed with monitors turned off (therefore covertly). | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:39:50 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Sun",
"Xieyang",
""
],
[
"Zheng",
"Yuanqing",
""
],
[
"Xi",
"Wei",
""
],
[
"Chen",
"Zuhao",
""
],
[
"Chen",
"Zhizhen",
""
],
[
"Hao",
"Han",
""
],
[
"Jiang",
"Zhiping",
""
],
[
"Zhong",
"Sheng",
""
]
] |
2506.21070 | Yikan Liu | Kai Yu, Zhiyuan Li, Yikan Liu | Inverse source problem with a posteriori interior measurements for space-time fractional diffusion equations | 14 pages, 2 figures, 2 tables | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper investigates an inverse source problem for space-time fractional diffusion equations from a posteriori interior measurements. The uniqueness result is established by the memory effect of fractional derivatives and the unique continuation property. For the numerical reconstruction, the inverse problem is reformulated as an optimization problem with the Tikhonov regularization. We use the Levenberg-Marquardt method to identity the unknown source from noisy measurements. Finally, we give some numerical examples to illustrate the efficiency and accuracy of the proposed algorithm. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:43:02 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yu",
"Kai",
""
],
[
"Li",
"Zhiyuan",
""
],
[
"Liu",
"Yikan",
""
]
] |
2506.21071 | Jingwei Wang | Jingwei Wang, Zai Zhang, Hao Qian, Chunjing Gan, Binbin Hu, Ziqi Liu, Zhiqiang Zhang, Jun Zhou, Bin Shi, Bo Dong | Enhancing LLM Tool Use with High-quality Instruction Data from Knowledge Graph | 20 pages, 12 figures | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | Teaching large language models (LLMs) to use tools is crucial for improving their problem-solving abilities and expanding their applications. However, effectively using tools is challenging because it requires a deep understanding of tool functionalities and user intentions. Previous methods relied mainly on LLMs to generate instruction data, but the quality of these data was often insufficient. In this paper, we propose a new method that uses knowledge graphs to generate high-quality instruction data for LLMs. Knowledge graphs are manually curated datasets rich in semantic information. We begin by extracting various query pathways from a given knowledge graph, which are transformed into a broad spectrum of user queries. We then translate the relationships between entities into actionable tools and parse the pathways of each query into detailed solution steps, thereby creating high-quality instruction data. Our experiments show that fine-tuning on just a small sample of this synthetic data can significantly improve the tool utilization and overall capabilities of LLMs. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:45:15 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Wang",
"Jingwei",
""
],
[
"Zhang",
"Zai",
""
],
[
"Qian",
"Hao",
""
],
[
"Gan",
"Chunjing",
""
],
[
"Hu",
"Binbin",
""
],
[
"Liu",
"Ziqi",
""
],
[
"Zhang",
"Zhiqiang",
""
],
[
"Zhou",
"Jun",
""
],
[
"Shi",
"Bin",
""
],
[
"Dong",
"Bo",
""
]
] |
2506.21072 | Carlos Jaime Barrios Hernandez | Carlos J Barrios (LIG, UIS, CITI), Yves Denneulin (LIG, Grenoble INP) | Bridding OT and PaaS in Edge-to-Cloud Continuum | null | Conf{\'e}rence Francophone d'Informatique en Parall{\'e}lisme, Architecture et Syst{\`e}me (COMPAS 2025), INRIA; UNIVERSITE DE BORDEAUX; CNRS, Jun 2025, BORDEAUX, France | null | null | cs.DC cs.PF | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Operational Technology Platform as a Service (OTPaaS) initiative provides a structured framework for the efficient management and storage of data. It ensures excellent response times while improving security, reliability, data and technology sovereignty, robustness, and energy efficiency, which are crucial for industrial transformation and data sovereignty. This paper illustrates successful deployment, adaptable application management, and various integration components catering to Edge and Cloud environments. It leverages the advantages of the Platform as a Service model and highlights key challenges that have been addressed for specific use cases. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:52:30 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Barrios",
"Carlos J",
"",
"LIG, UIS, CITI"
],
[
"Denneulin",
"Yves",
"",
"LIG, Grenoble INP"
]
] |
2506.21073 | Ilias Papalamprou | Ilias Papalamprou, Nikolaos Fotos, Nikolaos Chatzivasileiadis, Anna Angelogianni, Dimosthenis Masouros, Dimitrios Soudris | Post-Quantum and Blockchain-Based Attestation for Trusted FPGAs in B5G Networks | null | null | null | null | cs.AR | http://creativecommons.org/licenses/by/4.0/ | The advent of 5G and beyond has brought increased performance networks, facilitating the deployment of services closer to the user. To meet performance requirements such services require specialized hardware, such as Field Programmable Gate Arrays (FPGAs). However, FPGAs are often deployed in unprotected environments, leaving the user's applications vulnerable to multiple attacks. With the rise of quantum computing, which threatens the integrity of widely-used cryptographic algorithms, the need for a robust security infrastructure is even more crucial. In this paper we introduce a hybrid hardware-software solution utilizing remote attestation to securely configure FPGAs, while integrating Post-Quantum Cryptographic (PQC) algorithms for enhanced security. Additionally, to enable trustworthiness across the whole edge computing continuum, our solution integrates a blockchain infrastructure, ensuring the secure storage of any security evidence. We evaluate the proposed secure configuration process under different PQC algorithms in two FPGA families, showcasing only 2% overheard compared to the non PQC approach. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:58:22 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Papalamprou",
"Ilias",
""
],
[
"Fotos",
"Nikolaos",
""
],
[
"Chatzivasileiadis",
"Nikolaos",
""
],
[
"Angelogianni",
"Anna",
""
],
[
"Masouros",
"Dimosthenis",
""
],
[
"Soudris",
"Dimitrios",
""
]
] |
2506.21074 | Hankun Wang | Hankun Wang, Yiwei Guo, Chongtian Shao, Bohan Li, Xie Chen, Kai Yu | CodecSlime: Temporal Redundancy Compression of Neural Speech Codec via Dynamic Frame Rate | 16 pages, 5 figures, 9 tables | null | null | null | eess.AS cs.SD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Neural speech codecs have been widely used in audio compression and various downstream tasks. Current mainstream codecs are fixed-frame-rate (FFR), which allocate the same number of tokens to every equal-duration slice. However, speech is inherently non-uniform in temporal information density. As a result, many tokens are wasted on steady-state segments like long vowels and silences. To address this mismatch, we present CodecSlime, a plugin-style method for compressing temporal redundancy through supporting dynamic frame rate (DFR) on neural speech codecs for the first time. Our method is unsupervised and architecture-agnostic, combining two key innovations, ScheDFR and Melt-and-Cool, for adapting inference and training, respectively. When integrated into a typical VQ-GAN codec backbone and operating at 40 Hz DFR ($\approx$ 600 bps), the reconstruction WER of CodecSlime is reduced by up to 46% relative to conventional FFR baselines with the same model architecture and similar bitrates, while other metrics are also competitive. CodecSlime also enables flexible trade-offs between reconstruction quality and bitrate: a single model supports inference at multiple frame rates and consistently outperforms FFR models at the corresponding frame rates. Audio samples are available at https://acadarmeria.github.io/codecslime/. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 07:59:04 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Wang",
"Hankun",
""
],
[
"Guo",
"Yiwei",
""
],
[
"Shao",
"Chongtian",
""
],
[
"Li",
"Bohan",
""
],
[
"Chen",
"Xie",
""
],
[
"Yu",
"Kai",
""
]
] |
2506.21075 | Jiashun Guan | Jiashun Guan, Haoyang Hu, Tianfang Hao, Huimin Wang, Yunxiao Ren and Dixia Fan | Vortex-Induced Drag Forecast for Cylinder in Non-uniform Inflow | 6 pages, 7 figures | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | In this letter, a physics-based data-driven strategy is developed to predict vortex-induced drag on a circular cylinder under non-uniform inflow conditions - a prevalent issue for engineering applications at moderate Reynolds numbers. Traditional pressure-signal-based models exhibit limitations due to complex vortex dynamics coupled with non-uniform inflow. To address this issue, a modified fully connected neural network (FCNN) architecture is established that integrates upstream velocity measurements (serving as an inflow calibration) with pressure-signal-based inputs to enhance predictive capability (R^2 ~ 0 to 0.75). Direct numerical simulations (DNS) at Reynolds number Re = 4000 are implemented for model training and validation. Iterative optimizations are conducted to derive optimized input configurations of pressure sensor placements and velocity components at upstream locations. The optimized model achieves an R^2 score of 0.75 in forecasting high-amplitude drag coefficient fluctuations (C_d=0.2 - 1.2) within a future time window of one time unit. An exponential scaling between model performance and optimized pressure signal inputs is observed, and the predictive capability of sparsely distributed but optimized sensors is interpreted by the scaling. The optimized sensor placements correspond to the physical mechanism that the flow separation dynamics play a governing role in vortex-induced drag generation. This work advances machine learning applications in fluid-structure interaction systems, offering a scalable strategy for forecasting statistics in turbulent flows under real-world engineering conditions. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:01:50 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Guan",
"Jiashun",
""
],
[
"Hu",
"Haoyang",
""
],
[
"Hao",
"Tianfang",
""
],
[
"Wang",
"Huimin",
""
],
[
"Ren",
"Yunxiao",
""
],
[
"Fan",
"Dixia",
""
]
] |
2506.21076 | Hongyu Yan | Hongyu Yan, Kunming Luo, Weiyu Li, Yixun Liang, Shengming Li, Jingwei Huang, Chunchao Guo, Ping Tan | PoseMaster: Generating 3D Characters in Arbitrary Poses from a Single Image | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | 3D characters play a crucial role in our daily entertainment. To improve the efficiency of 3D character modeling, recent image-based methods use two separate models to achieve pose standardization and 3D reconstruction of the A-pose character. However, these methods are prone to generating distorted and degraded images in the pose standardization stage due to self-occlusion and viewpoints, which further affects the geometric quality of the subsequent reconstruction process. To tackle these problems, we propose PoseMaster, an end-to-end controllable 3D character generation framework. Specifically, we unify pose transformation and 3D character generation into a flow-based 3D native generation framework. To achieve accurate arbitrary-pose control, we propose to leverage the 3D body bones existing in the skeleton of an animatable character as the pose condition. Furthermore, considering the specificity of multi-condition control, we randomly empty the pose condition and the image condition during training to improve the effectiveness and generalizability of pose control. Finally, we create a high-quality pose-control dataset derived from realistic character animation data to make the model learning the implicit relationships between skeleton and skinning weights. Extensive experiments show that PoseMaster outperforms current state-of-the-art techniques in both qualitative and quantitative evaluations for A-pose character generation while demonstrating its powerful ability to achieve precise control for arbitrary poses. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:03:14 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yan",
"Hongyu",
""
],
[
"Luo",
"Kunming",
""
],
[
"Li",
"Weiyu",
""
],
[
"Liang",
"Yixun",
""
],
[
"Li",
"Shengming",
""
],
[
"Huang",
"Jingwei",
""
],
[
"Guo",
"Chunchao",
""
],
[
"Tan",
"Ping",
""
]
] |
2506.21077 | Kaicheng Zhang | Kaicheng Zhang, Shida Xu, Yining Ding, Xianwen Kong, Sen Wang | CURL-SLAM: Continuous and Compact LiDAR Mapping | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper studies 3D LiDAR mapping with a focus on developing an updatable and localizable map representation that enables continuity, compactness and consistency in 3D maps. Traditional LiDAR Simultaneous Localization and Mapping (SLAM) systems often rely on 3D point cloud maps, which typically require extensive storage to preserve structural details in large-scale environments. In this paper, we propose a novel paradigm for LiDAR SLAM by leveraging the Continuous and Ultra-compact Representation of LiDAR (CURL) introduced in [1]. Our proposed LiDAR mapping approach, CURL-SLAM, produces compact 3D maps capable of continuous reconstruction at variable densities using CURL's spherical harmonics implicit encoding, and achieves global map consistency after loop closure. Unlike popular Iterative Closest Point (ICP)-based LiDAR odometry techniques, CURL-SLAM formulates LiDAR pose estimation as a unique optimization problem tailored for CURL and extends it to local Bundle Adjustment (BA), enabling simultaneous pose refinement and map correction. Experimental results demonstrate that CURL-SLAM achieves state-of-the-art 3D mapping quality and competitive LiDAR trajectory accuracy, delivering sensor-rate real-time performance (10 Hz) on a CPU. We will release the CURL-SLAM implementation to the community. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:05:30 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Zhang",
"Kaicheng",
""
],
[
"Xu",
"Shida",
""
],
[
"Ding",
"Yining",
""
],
[
"Kong",
"Xianwen",
""
],
[
"Wang",
"Sen",
""
]
] |
2506.21078 | Tian Han | Tian Han, Shalanika Dayarathna, Rajitha Senanayake, Peter Smith, Aryan Kaushik, Alain Mourad, Richard A. Stirling-Gallacher, Jamie Evans | Constant Modulus Waveforms for IoT-Centric Integrated Sensing and Communications | Submitted for publication to IEEE Communications Standards Magazine | null | null | null | cs.IT eess.SP math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Integrated sensing and communications (ISAC) is considered a key enabler to support application scenarios such as the Internet-of-Things (IoT) in which both communications and sensing play significant roles. Multi-carrier waveforms, such as orthogonal frequency division multiplexing (OFDM), have been considered as good candidates for ISAC due to their high communications data rate and good time bandwidth property for sensing. Nevertheless, their high peak-to-average-power-ratio (PAPR) values lead to either performance degradation or an increase in system complexity. This can make OFDM unsuitable for IoT applications with insufficient resources in terms of power, system complexity, hardware size or cost. This article provides IoT-centric constant modulus waveform designs that leverage the advantage of unit PAPR and thus are more suitable in resource-limited scenarios. More specifically, several single-carrier frequency and/or phase-modulated waveforms are considered. A comprehensive discussion on their radar sensing and communications performance is conducted based on performance metrics, including the radar ambiguity function, the bandwidth property, the data rate, and the communications receiver complexity. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:06:24 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Han",
"Tian",
""
],
[
"Dayarathna",
"Shalanika",
""
],
[
"Senanayake",
"Rajitha",
""
],
[
"Smith",
"Peter",
""
],
[
"Kaushik",
"Aryan",
""
],
[
"Mourad",
"Alain",
""
],
[
"Stirling-Gallacher",
"Richard A.",
""
],
[
"Evans",
"Jamie",
""
]
] |
2506.21079 | YANN KERZREHO | Yann Kerzreho (ENS Paris Saclay) | Homogenization of Multi-agent Learning Dynamics in Finite-state Markov Games | null | null | null | null | stat.ML cs.LG math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces a new approach for approximating the learning dynamics of multiple reinforcement learning (RL) agents interacting in a finite-state Markov game. The idea is to rescale the learning process by simultaneously reducing the learning rate and increasing the update frequency, effectively treating the agent's parameters as a slow-evolving variable influenced by the fast-mixing game state. Under mild assumptions-ergodicity of the state process and continuity of the updates-we prove the convergence of this rescaled process to an ordinary differential equation (ODE). This ODE provides a tractable, deterministic approximation of the agent's learning dynamics. An implementation of the framework is available at\,: https://github.com/yannKerzreho/MarkovGameApproximation | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:08:49 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Kerzreho",
"Yann",
"",
"ENS Paris Saclay"
]
] |
2506.21080 | Sanjoy Chowdhury | Sanjoy Chowdhury, Subrata Biswas, Sayan Nag, Tushar Nagarajan, Calvin Murdock, Ishwarya Ananthabhotla, Yijun Qian, Vamsi Krishna Ithapu, Dinesh Manocha, Ruohan Gao | EgoAdapt: Adaptive Multisensory Distillation and Policy Learning for Efficient Egocentric Perception | Accepted at ICCV 2025 | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Modern perception models, particularly those designed for multisensory egocentric tasks, have achieved remarkable performance but often come with substantial computational costs. These high demands pose challenges for real-world deployment, especially in resource-constrained environments. In this paper, we introduce EgoAdapt, a framework that adaptively performs cross-modal distillation and policy learning to enable efficient inference across different egocentric perception tasks, including egocentric action recognition, active speaker localization, and behavior anticipation. Our proposed policy module is adaptable to task-specific action spaces, making it broadly applicable. Experimental results on three challenging egocentric datasets EPIC-Kitchens, EasyCom, and Aria Everyday Activities demonstrate that our method significantly enhances efficiency, reducing GMACs by up to 89.09%, parameters up to 82.02%, and energy up to 9.6x, while still on-par and in many cases outperforming, the performance of corresponding state-of-the-art models. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:09:16 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Chowdhury",
"Sanjoy",
""
],
[
"Biswas",
"Subrata",
""
],
[
"Nag",
"Sayan",
""
],
[
"Nagarajan",
"Tushar",
""
],
[
"Murdock",
"Calvin",
""
],
[
"Ananthabhotla",
"Ishwarya",
""
],
[
"Qian",
"Yijun",
""
],
[
"Ithapu",
"Vamsi Krishna",
""
],
[
"Manocha",
"Dinesh",
""
],
[
"Gao",
"Ruohan",
""
]
] |
2506.21081 | Germain BOURCIER | Germain Bourcier (LAAS-MOST), Stephane Balac (IRMAR), Julien Lumeau (RCMO, FRESNEL), Antonin Moreau (FRESNEL), Thomas Bunel (PhLAM), Arnaud Mussot (PhLAM), Matteo Conforti (Photonique, IRCICA, PhLAM), Vincent Crozatier, Olivier Llopis (LAAS-MOST), Arnaud Fernandez (LAAS-MOST) | Laser self-injection locking to fiber Fabry-Perot resonator for frequency comb generation | null | null | null | null | physics.optics | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study demonstrates that self-injection locking (SIL) of a distributed feedback (DFB) laser to a high-Q fiber Fabry Perot (FFP) resonator, fabricated with highly nonlinear fiber, allows optical frequency combs (OFC) generation with a laser power as low as 100 mW. More precisely, cavity soliton (CS) regime has been observed in this configuration, along with other types of combs. The laser stabilization using SIL is described. Then the system's behavior is analyzed through modeling the laser's dynamics and comparing the model results to experimental tuning curve measurements. Our findings highlight the critical role of the initial phase of the fiber link between laser and FFP in determining the stability and effectiveness of the locking process. We explore the dynamics of the nonlinear SIL process while varying the laser current, revealing the transition from modulation instability to chaotic comb states, and eventually to soliton formation as the system moves from an effective blue-detuned to an effective red-detuned regime. Notably, the inclusion of self-phase modulation (SPM) in the SIL model predicts accessibility of the narrow soliton existence range. These results highlight the potential of SIL in FFP resonators for low-power, stable OFC generation, offering a promising path forward for practical applications. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:09:37 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Bourcier",
"Germain",
"",
"LAAS-MOST"
],
[
"Balac",
"Stephane",
"",
"IRMAR"
],
[
"Lumeau",
"Julien",
"",
"RCMO, FRESNEL"
],
[
"Moreau",
"Antonin",
"",
"FRESNEL"
],
[
"Bunel",
"Thomas",
"",
"PhLAM"
],
[
"Mussot",
"Arnaud",
"",
"PhLAM"
],
[
"Conforti",
"Matteo",
"",
"Photonique, IRCICA, PhLAM"
],
[
"Crozatier",
"Vincent",
"",
"LAAS-MOST"
],
[
"Llopis",
"Olivier",
"",
"LAAS-MOST"
],
[
"Fernandez",
"Arnaud",
"",
"LAAS-MOST"
]
] |
2506.21082 | Fei Ren | Fei Ren | Coherent six-functor formalisms: Pro vs Solid | 14 pages. Comments welcome! | null | null | null | math.AG math.GN | http://creativecommons.org/licenses/by/4.0/ | In the classical theory for coherent sheaves, the only missing piece in the Grothendieck six-functor formalism picture is $j_!$ for an open immersion $j$. Towards fixing this gap,classically Deligne proposed a construction of $j_!$ by extending the sheaf class to pro sheaves, while Clausen-Scholze provided another solution by extending the sheaf class to solid modules. In this work, we prove that Deligne's construction coincides with the Clausen-Scholze construction via a natural functor, whose restriction to the full subcategory of Mittag-Leffler pro-systems is fully faithful. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:16:51 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Ren",
"Fei",
""
]
] |
2506.21083 | Yash Ingole | Yash B. Ingole, K.V. Somwanshi, S.R. Bhoyar and S.N. Bayaskar | Exploring Dark Energy via Non-Minimal Coupling in $f(Q,L_m)$ Gravity with Gong-Zhang Parameterization | 19 pages and 15 figures | null | null | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | In this study, we investigate the late-time accelerated expansion of the universe within the framework of non-minimally coupled $f(Q,L_m)$ gravity, where $Q$ is the non-metricity scalar and $L_m$ is the matter Lagrangian. We derive modified Friedmann equations in a flat FLRW background and employ the \textit{Gong-Zhang} parameterization for the DE equation of state (EoS), allowing an analytical form of the Hubble parameter $H(z)$. The model parameters are constrained using recent Cosmic Chronometers (CC) and Pantheon+SH0ES Type Ia supernova datasets through MCMC-based chi-squared minimization. We analyze various cosmological quantities including the deceleration parameter, EoS, jerk, snap, lerk, and diagnostic tools such as $Om(z)$ and the statefinder pair $(r,s)$. Our findings indicate a viable transition from deceleration to acceleration and reveal a quintessence-to-phantom-like evolution of dark energy. Furthermore, energy conditions are examined, showing a violation of the strong energy condition, consistent with current cosmic acceleration. The results establish that the $f(Q,L_m) $ framework with non-minimal coupling and parameterized EoS provides a compelling alternative to $\Lambda$CDM in describing cosmic acceleration. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:17:15 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Ingole",
"Yash B.",
""
],
[
"Somwanshi",
"K. V.",
""
],
[
"Bhoyar",
"S. R.",
""
],
[
"Bayaskar",
"S. N.",
""
]
] |
2506.21084 | Pablo Concha-Vega | Pablo Concha-Vega (AMU, LIS), K\'evin Perrot (AMU, LIS) | Timed Prediction Problem for Sandpile Models | null | null | null | null | cs.CC nlin.CG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the computational complexity of the timed prediction problem in two-dimensional sandpile models. This question refines the classical prediction problem, which asks whether a cell q will eventually become unstable after adding a grain at cell p from a given configuration. The prediction problem has been shown to be P-complete in several settings, including for subsets of the Moore neighborhood, but its complexity for the von Neumann neighborhood remains open. In a previous work, we provided a complete characterization of crossover gates (a key to the implementation of non-planar monotone circuits) for these small neighborhoods, leading to P-completeness proofs with only 4 and 5 neighbors among the eight adjancent cells. In this paper, we introduce the timed setting, where the goal is to determine whether cell q becomes unstable exactly at time t. We distinguish several cases: some neighborhoods support complete timed toolkits (including timed crossover gates) and exhibit P-completeness; others admit timed crossovers but suffer from synchronization issues; planar neighborhoods provably do not admit any timed crossover; and finally, for some remaining neighborhoods, we conjecture that no timed crossover is possible. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:19:31 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Concha-Vega",
"Pablo",
"",
"AMU, LIS"
],
[
"Perrot",
"Kévin",
"",
"AMU, LIS"
]
] |
2506.21085 | Yangzhe Peng | Yangzhe Peng, Kaiyuan Gao, Liang He, Yuheng Cong, Haiguang Liu, Kun He, Lijun Wu | CovDocker: Benchmarking Covalent Drug Design with Tasks, Datasets, and Solutions | Accepted to KDD 2025 Research Track | null | 10.1145/3711896.3736896 | null | q-bio.BM cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Molecular docking plays a crucial role in predicting the binding mode of ligands to target proteins, and covalent interactions, which involve the formation of a covalent bond between the ligand and the target, are particularly valuable due to their strong, enduring binding nature. However, most existing docking methods and deep learning approaches hardly account for the formation of covalent bonds and the associated structural changes. To address this gap, we introduce a comprehensive benchmark for covalent docking, CovDocker, which is designed to better capture the complexities of covalent binding. We decompose the covalent docking process into three main tasks: reactive location prediction, covalent reaction prediction, and covalent docking. By adapting state-of-the-art models, such as Uni-Mol and Chemformer, we establish baseline performances and demonstrate the effectiveness of the benchmark in accurately predicting interaction sites and modeling the molecular transformations involved in covalent binding. These results confirm the role of the benchmark as a rigorous framework for advancing research in covalent drug design. It underscores the potential of data-driven approaches to accelerate the discovery of selective covalent inhibitors and addresses critical challenges in therapeutic development. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:28:07 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Peng",
"Yangzhe",
""
],
[
"Gao",
"Kaiyuan",
""
],
[
"He",
"Liang",
""
],
[
"Cong",
"Yuheng",
""
],
[
"Liu",
"Haiguang",
""
],
[
"He",
"Kun",
""
],
[
"Wu",
"Lijun",
""
]
] |
2506.21086 | Guillem Cort\`es-Sebasti\`a | Guillem Cort\`es-Sebasti\`a and Benjamin Martin and Emilio Molina and Xavier Serra and Romain Hennequin | PeakNetFP: Peak-based Neural Audio Fingerprinting Robust to Extreme Time Stretching | Accepted at ISMIR 2025 | null | null | null | cs.SD cs.IR eess.AS | http://creativecommons.org/licenses/by/4.0/ | This work introduces PeakNetFP, the first neural audio fingerprinting (AFP) system designed specifically around spectral peaks. This novel system is designed to leverage the sparse spectral coordinates typically computed by traditional peak-based AFP methods. PeakNetFP performs hierarchical point feature extraction techniques similar to the computer vision model PointNet++, and is trained using contrastive learning like in the state-of-the-art deep learning AFP, NeuralFP. This combination allows PeakNetFP to outperform conventional AFP systems and achieves comparable performance to NeuralFP when handling challenging time-stretched audio data. In extensive evaluation, PeakNetFP maintains a Top-1 hit rate of over 90% for stretching factors ranging from 50% to 200%. Moreover, PeakNetFP offers significant efficiency advantages: compared to NeuralFP, it has 100 times fewer parameters and uses 11 times smaller input data. These features make PeakNetFP a lightweight and efficient solution for AFP tasks where time stretching is involved. Overall, this system represents a promising direction for future AFP technologies, as it successfully merges the lightweight nature of peak-based AFP with the adaptability and pattern recognition capabilities of neural network-based approaches, paving the way for more scalable and efficient solutions in the field. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:29:48 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Cortès-Sebastià",
"Guillem",
""
],
[
"Martin",
"Benjamin",
""
],
[
"Molina",
"Emilio",
""
],
[
"Serra",
"Xavier",
""
],
[
"Hennequin",
"Romain",
""
]
] |
2506.21087 | Fabien Panloup | Mohamed Alfaki Aboubacrine Assadeck (LAREMA), Fabien Panloup (LAREMA) | Quasi-Stationary Distributions of Interacting Dynamical Systems and their approximation | null | null | null | null | math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In \cite{BCP}, the authors built and studied an algorithm based on the (self)-interaction of a dynamics with its occupation measure to approximate Quasi-Stationary Distributions (QSD) of general Markov chains conditioned to stay in a compact set. In this paper, we propose to tackle the case of McKean-Vlasov-type dynamics, \emph{i.e.} of dynamics interacting with their marginal distribution (conditioned to not be killed). In this non-linear setting, we are able to exhibit some conditions which guarantee that weak limits of these sequences of random measures are QSDs of the given dynamics. We also prove tightness results in the non-compact case. These general conditions are then applied to Euler schemes of McKean-Vlasov SDEs and in the compact case, the behavior of these QSDs when the step $h$ goes to $0$ is investigated. Our results also allow to consider some examples in the non-compact case and some new tightness criterions are also provided in this setting. Finally, we illustrate our theoretical results with several simulations. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:30:30 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Assadeck",
"Mohamed Alfaki Aboubacrine",
"",
"LAREMA"
],
[
"Panloup",
"Fabien",
"",
"LAREMA"
]
] |
2506.21088 | Ramsamoj Kewat | Ramsamoj Kewat and Nirmal Ganguli | Nonsymmorphic symmetry-enforced hourglass fermions and Rashba-Dresselhaus interaction in BiInO$_3$ | null | null | null | null | cond-mat.mtrl-sci cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | In this study, we investigate the spin texture of the hourglass fermions band network in BiInO$_3$ using density functional theory (DFT) and symmetry analysis. Hourglass fermions are of interest in spintronics due to their unique and robust band structure, as well as their potential applications in novel electronic devices. BiInO$_3$ exhibits non-symmorphic crystal symmetries, such as glide reflection and glide rotational symmetry, influencing its electronic properties. Through symmetry analysis, we explore the band crossings and spin textures along specific high-symmetry paths in the Brillouin zone. Our results reveal a fascinating hourglass-shaped band dispersion and spin polarisation governed by symmetry operations and spin-orbit interaction. We analyse the spin-splitting mechanisms, including Dresselhaus and Rashba spin-orbit interactions, and suggest potential applications for spin-based devices. This study sheds light on the role of symmetry in crystals for fascinating spin properties of hourglass fermions in non-symmorphic materials, offering insights for future developments in spintronics. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:30:35 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Kewat",
"Ramsamoj",
""
],
[
"Ganguli",
"Nirmal",
""
]
] |
2506.21089 | Jingkai Quan | Jingkai Quan, Nikita Rybin, Matthias Scheffler, Christian Carbogno | Efficient Band Structure Unfolding with Atomic-centered Orbitals: General Theory and Application | null | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Band structure unfolding is a key technique for analyzing and simplifying the electronic band structure of large, internally distorted supercells that break the primitive cell's translational symmetry. In this work, we present an efficient band unfolding method for atomic orbital (AO) basis sets that explicitly accounts for both the non-orthogonality of atomic orbitals and their atom-centered nature. Unlike existing approaches that typically rely on a plane-wave representation of the (semi-)valence states, we here derive analytical expressions that recasts the primitive cell translational operator and the associated Bloch-functions in the supercell AO basis. In turn, this enables the accurate and efficient unfolding of conduction, valence, and core states in all-electron codes, as demonstrated by our implementation in the all-electron ab initio simulation package FHI-aims, which employs numeric atom-centered orbitals. We explicitly demonstrate the capability of running large-scale unfolding calculations for systems with thousands of atoms and showcase the importance of this technique for computing temperature-dependent spectral functions in strongly anharmonic materials using CuI as example. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:31:35 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Quan",
"Jingkai",
""
],
[
"Rybin",
"Nikita",
""
],
[
"Scheffler",
"Matthias",
""
],
[
"Carbogno",
"Christian",
""
]
] |
2506.21090 | Wanying Ge | Wanying Ge, Xin Wang, Xuechen Liu, Junichi Yamagishi | Post-training for Deepfake Speech Detection | null | null | null | null | eess.AS | http://creativecommons.org/licenses/by/4.0/ | We introduce a post-training approach that adapts self-supervised learning (SSL) models for deepfake speech detection by bridging the gap between general pre-training and domain-specific fine-tuning. We present AntiDeepfake models, a series of post-trained models developed using a large-scale multilingual speech dataset containing over 56,000 hours of genuine speech and 18,000 hours of speech with various artifacts in over one hundred languages. Experimental results show that the post-trained models already exhibit strong robustness and generalization to unseen deepfake speech. When they are further fine-tuned on the Deepfake-Eval-2024 dataset, these models consistently surpass existing state-of-the-art detectors that do not leverage post-training. Model checkpoints and source code are available online. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:34:19 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Ge",
"Wanying",
""
],
[
"Wang",
"Xin",
""
],
[
"Liu",
"Xuechen",
""
],
[
"Yamagishi",
"Junichi",
""
]
] |
2506.21091 | Mahmoud Tahmasebi Mr | Mahmoud Tahmasebi, Saif Huq, Kevin Meehan, Marion McAfee | ESMStereo: Enhanced ShuffleMixer Disparity Upsampling for Real-Time and Accurate Stereo Matching | Under peer review | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Stereo matching has become an increasingly important component of modern autonomous systems. Developing deep learning-based stereo matching models that deliver high accuracy while operating in real-time continues to be a major challenge in computer vision. In the domain of cost-volume-based stereo matching, accurate disparity estimation depends heavily on large-scale cost volumes. However, such large volumes store substantial redundant information and also require computationally intensive aggregation units for processing and regression, making real-time performance unattainable. Conversely, small-scale cost volumes followed by lightweight aggregation units provide a promising route for real-time performance, but lack sufficient information to ensure highly accurate disparity estimation. To address this challenge, we propose the Enhanced Shuffle Mixer (ESM) to mitigate information loss associated with small-scale cost volumes. ESM restores critical details by integrating primary features into the disparity upsampling unit. It quickly extracts features from the initial disparity estimation and fuses them with image features. These features are mixed by shuffling and layer splitting then refined through a compact feature-guided hourglass network to recover more detailed scene geometry. The ESM focuses on local contextual connectivity with a large receptive field and low computational cost, leading to the reconstruction of a highly accurate disparity map at real-time. The compact version of ESMStereo achieves an inference speed of 116 FPS on high-end GPUs and 91 FPS on the AGX Orin. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:34:51 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Tahmasebi",
"Mahmoud",
""
],
[
"Huq",
"Saif",
""
],
[
"Meehan",
"Kevin",
""
],
[
"McAfee",
"Marion",
""
]
] |
2506.21092 | Alexander Stokes | Alexander Stokes, Tomoyuki Takenawa and Adrian Stefan Carstea | On the geometry of a 4-dimensional extension of a $q$-Painlev\'e I equation with symmetry type $A_1^{(1)}$ | 25 pages | null | null | null | nlin.SI math-ph math.AG math.DS math.MP | http://creativecommons.org/licenses/by/4.0/ | We present a geometric study of a four-dimensional integrable discrete dynamical system which extends the autonomous form of a $q$-Painlev\'e I equation with symmetry of type $A_1^{(1)}$. By resolution of singularities it is lifted to a pseudo-automorphism of a rational variety obtained from $({\mathbb P}^1)^{\times 4}$ by blowing up along 28 subvarieties and we use this to establish its integrability in terms of conserved quantities and degree growth. We embed this rational variety into a family which admits an action of the extended affine Weyl group $\widetilde{W}(A_1^{(1)})\times \widetilde{W}(A_1^{(1)})$ by pseudo-isomorphisms. We use this to construct two 4-dimensional analogues of $q$-Painlev\'e equations, one of which is a deautonomisation of the original autonomous integrable map. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:39:55 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Stokes",
"Alexander",
""
],
[
"Takenawa",
"Tomoyuki",
""
],
[
"Carstea",
"Adrian Stefan",
""
]
] |
2506.21093 | Li Fan | Li Fan, Peng Wang, Jing Yang, Cong Shen | Chain-of-Thought Enhanced Shallow Transformers for Wireless Symbol Detection | null | null | null | null | cs.LG cs.IT eess.SP math.IT stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Transformers have shown potential in solving wireless communication problems, particularly via in-context learning (ICL), where models adapt to new tasks through prompts without requiring model updates. However, prior ICL-based Transformer models rely on deep architectures with many layers to achieve satisfactory performance, resulting in substantial storage and computational costs. In this work, we propose CHain Of thOught Symbol dEtection (CHOOSE), a CoT-enhanced shallow Transformer framework for wireless symbol detection. By introducing autoregressive latent reasoning steps within the hidden space, CHOOSE significantly improves the reasoning capacity of shallow models (1-2 layers) without increasing model depth. This design enables lightweight Transformers to achieve detection performance comparable to much deeper models, making them well-suited for deployment on resource-constrained mobile devices. Experimental results demonstrate that our approach outperforms conventional shallow Transformers and achieves performance comparable to that of deep Transformers, while maintaining storage and computational efficiency. This represents a promising direction for implementing Transformer-based algorithms in wireless receivers with limited computational resources. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:41:45 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Fan",
"Li",
""
],
[
"Wang",
"Peng",
""
],
[
"Yang",
"Jing",
""
],
[
"Shen",
"Cong",
""
]
] |
2506.21094 | Chonfai Kam | Chon-Fai Kam and En-Jui Kuo | Quantum Supremacy through Fock State $q$ boson Sampling with Transmon Qubits | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Transmon qubits have traditionally been regarded as limited to random circuit sampling, incapable of performing Fock state boson sampling, a problem known to be classically intractable. This work challenges that assumption by introducing $q$ boson Fock state sampling, a variant in which transmon qubits can operate effectively. Through direct mapping to the $q$ boson formalism, we demonstrate that transmons possess the capability to achieve quantum supremacy in $q$ boson sampling tasks. This finding expands the potential applications of transmon-based quantum processors and paves the way for new avenues in quantum computation. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:41:45 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Kam",
"Chon-Fai",
""
],
[
"Kuo",
"En-Jui",
""
]
] |
2506.21095 | Mattia Cerrato | Xenia Heilmann, Luca Corbucci, Mattia Cerrato, Anna Monreale | FeDa4Fair: Client-Level Federated Datasets for Fairness Evaluation | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Federated Learning (FL) enables collaborative model training across multiple clients without sharing clients' private data. However, fairness remains a key concern, as biases in local clients' datasets can impact the entire federated system. Heterogeneous data distributions across clients may lead to models that are fairer for some clients than others. Although several fairness-enhancing solutions are present in the literature, most focus on mitigating bias for a single sensitive attribute, typically binary, overlooking the diverse and sometimes conflicting fairness needs of different clients. This limited perspective can limit the effectiveness of fairness interventions for the different clients. To support more robust and reproducible fairness research in FL, we aim to enable a consistent benchmarking of fairness-aware FL methods at both the global and client levels. In this paper, we contribute in three ways: (1) We introduce FeDa4Fair, a library to generate tabular datasets tailored to evaluating fair FL methods under heterogeneous client bias; (2) we release four bias-heterogeneous datasets and corresponding benchmarks to compare fairness mitigation methods in a controlled environment; (3) we provide ready-to-use functions for evaluating fairness outcomes for these datasets. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:43:12 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Heilmann",
"Xenia",
""
],
[
"Corbucci",
"Luca",
""
],
[
"Cerrato",
"Mattia",
""
],
[
"Monreale",
"Anna",
""
]
] |
2506.21096 | Kang He | Kang He, Yuzhe Ding. Haining Wang, Fei Li, Chong Teng, Donghong Ji | DALR: Dual-level Alignment Learning for Multimodal Sentence Representation Learning | Accepted by ACL 2025 Findings | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Previous multimodal sentence representation learning methods have achieved impressive performance. However, most approaches focus on aligning images and text at a coarse level, facing two critical challenges:cross-modal misalignment bias and intra-modal semantic divergence, which significantly degrade sentence representation quality. To address these challenges, we propose DALR (Dual-level Alignment Learning for Multimodal Sentence Representation). For cross-modal alignment, we propose a consistency learning module that softens negative samples and utilizes semantic similarity from an auxiliary task to achieve fine-grained cross-modal alignment. Additionally, we contend that sentence relationships go beyond binary positive-negative labels, exhibiting a more intricate ranking structure. To better capture these relationships and enhance representation quality, we integrate ranking distillation with global intra-modal alignment learning. Comprehensive experiments on semantic textual similarity (STS) and transfer (TR) tasks validate the effectiveness of our approach, consistently demonstrating its superiority over state-of-the-art baselines. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:45:14 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"He",
"Kang",
""
],
[
"Wang",
"Yuzhe Ding. Haining",
""
],
[
"Li",
"Fei",
""
],
[
"Teng",
"Chong",
""
],
[
"Ji",
"Donghong",
""
]
] |
2506.21097 | George Koutsoumbas | George Koutsoumbas, Andri Machattou, Eleftherios Papantonopoulos | Formation of Bound States in Quintessence Alternative Theories | 11 pages | null | null | null | gr-qc hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the formation and behaviour of bound states formed outside the horizon of a black hole in the presence of quintessence matter. Calculating the Regge and Wheeler potential for general metric function, we find that the presence of quintessence influences significantly the metric function and the Hawking temperature. We show that large black holes radiate less in the presence of large quintessence matter and it seems to live longer, while small black holes radiate more. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:45:41 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Koutsoumbas",
"George",
""
],
[
"Machattou",
"Andri",
""
],
[
"Papantonopoulos",
"Eleftherios",
""
]
] |
2506.21098 | Qinwen Chen | Qinwen Chen, Wenbiao Tao, Zhiwei Zhu, Mingfan Xi, Liangzhong Guo, Yuan Wang, Wei Wang, Yunshi Lan | ComRAG: Retrieval-Augmented Generation with Dynamic Vector Stores for Real-time Community Question Answering in Industry | 7 pages, 4 figures. Accepted at ACL 2025 Industry Track | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Community Question Answering (CQA) platforms can be deemed as important knowledge bases in community, but effectively leveraging historical interactions and domain knowledge in real-time remains a challenge. Existing methods often underutilize external knowledge, fail to incorporate dynamic historical QA context, or lack memory mechanisms suited for industrial deployment. We propose ComRAG, a retrieval-augmented generation framework for real-time industrial CQA that integrates static knowledge with dynamic historical QA pairs via a centroid-based memory mechanism designed for retrieval, generation, and efficient storage. Evaluated on three industrial CQA datasets, ComRAG consistently outperforms all baselines--achieving up to 25.9% improvement in vector similarity, reducing latency by 8.7% to 23.3%, and lowering chunk growth from 20.23% to 2.06% over iterations. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:48:16 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Chen",
"Qinwen",
""
],
[
"Tao",
"Wenbiao",
""
],
[
"Zhu",
"Zhiwei",
""
],
[
"Xi",
"Mingfan",
""
],
[
"Guo",
"Liangzhong",
""
],
[
"Wang",
"Yuan",
""
],
[
"Wang",
"Wei",
""
],
[
"Lan",
"Yunshi",
""
]
] |
2506.21099 | Ali Chamseddine | Ali H. Chamseddine and J\"urg Fr\"ohlich | On the Dark-Energy Enigma | 15 pages | null | null | null | gr-qc astro-ph.CO hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a model that offers an explanation for the presence of (Dark Matter and) Dark Energy in the universe. A key idea is to express the volume form of the Lorentzian metric on space-time in terms of a positive function of a new scalar field multiplying a certain four-form given by the wedge product of the differential of the mimetic scalar field and a certain closed three-form. An ansatz for this three-form related to one commonly used to determine the winding number of a map from a three-dimensional hypersurface to a three-sphere is discussed. An action functional depending on the space-time metric, the new scalar field, the mimetic scalar and the three-form is proposed, and the field equations are derived. Special solutions of these equations for a Friedmann-Lema\^itre universe are presented. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:53:02 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Chamseddine",
"Ali H.",
""
],
[
"Fröhlich",
"Jürg",
""
]
] |
2506.21100 | Vasilis Sarafidis Professor | Nektarios Aslanidis, Aurelio Bariviera, George Kapetanios, Vasilis Sarafidis | Heterogeneous Exposures to Systematic and Idiosyncratic Risk across Crypto Assets: A Divide-and-Conquer Approach | null | null | null | null | econ.EM q-fin.RM | http://creativecommons.org/licenses/by/4.0/ | This paper analyzes realized return behavior across a broad set of crypto assets by estimating heterogeneous exposures to idiosyncratic and systematic risk. A key challenge arises from the latent nature of broader economy-wide risk sources: macro-financial proxies are unavailable at high-frequencies, while the abundance of low-frequency candidates offers limited guidance on empirical relevance. To address this, we develop a two-stage ``divide-and-conquer'' approach. The first stage estimates exposures to high-frequency idiosyncratic and market risk only, using asset-level IV regressions. The second stage identifies latent economy-wide factors by extracting the leading principal component from the model residuals and mapping it to lower-frequency macro-financial uncertainty and sentiment-based indicators via high-dimensional variable selection. Structured patterns of heterogeneity in exposures are uncovered using Mean Group estimators across asset categories. The method is applied to a broad sample of crypto assets, covering more than 80% of total market capitalization. We document short-term mean reversion and significant average exposures to idiosyncratic volatility and illiquidity. Green and DeFi assets are, on average, more exposed to market-level and economy-wide risk than their non-Green and non-DeFi counterparts. By contrast, stablecoins are less exposed to idiosyncratic, market-level, and economy-wide risk factors relative to non-stablecoins. At a conceptual level, our study develops a coherent framework for isolating distinct layers of risk in crypto markets. Empirically, it sheds light on how return sensitivities vary across digital asset categories -- insights that are important for both portfolio design and regulatory oversight. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:54:03 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Aslanidis",
"Nektarios",
""
],
[
"Bariviera",
"Aurelio",
""
],
[
"Kapetanios",
"George",
""
],
[
"Sarafidis",
"Vasilis",
""
]
] |
2506.21101 | Caoshuo Li | Caoshuo Li, Zengmao Ding, Xiaobin Hu, Bang Li, Donghao Luo, AndyPian Wu, Chaoyang Wang, Chengjie Wang, Taisong Jin, SevenShu, Yunsheng Wu, Yongge Liu, Rongrong Ji | OracleFusion: Assisting the Decipherment of Oracle Bone Script with Structurally Constrained Semantic Typography | Accepted to ICCV 2025 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As one of the earliest ancient languages, Oracle Bone Script (OBS) encapsulates the cultural records and intellectual expressions of ancient civilizations. Despite the discovery of approximately 4,500 OBS characters, only about 1,600 have been deciphered. The remaining undeciphered ones, with their complex structure and abstract imagery, pose significant challenges for interpretation. To address these challenges, this paper proposes a novel two-stage semantic typography framework, named OracleFusion. In the first stage, this approach leverages the Multimodal Large Language Model (MLLM) with enhanced Spatial Awareness Reasoning (SAR) to analyze the glyph structure of the OBS character and perform visual localization of key components. In the second stage, we introduce Oracle Structural Vector Fusion (OSVF), incorporating glyph structure constraints and glyph maintenance constraints to ensure the accurate generation of semantically enriched vector fonts. This approach preserves the objective integrity of the glyph structure, offering visually enhanced representations that assist experts in deciphering OBS. Extensive qualitative and quantitative experiments demonstrate that OracleFusion outperforms state-of-the-art baseline models in terms of semantics, visual appeal, and glyph maintenance, significantly enhancing both readability and aesthetic quality. Furthermore, OracleFusion provides expert-like insights on unseen oracle characters, making it a valuable tool for advancing the decipherment of OBS. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:56:07 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Li",
"Caoshuo",
""
],
[
"Ding",
"Zengmao",
""
],
[
"Hu",
"Xiaobin",
""
],
[
"Li",
"Bang",
""
],
[
"Luo",
"Donghao",
""
],
[
"Wu",
"AndyPian",
""
],
[
"Wang",
"Chaoyang",
""
],
[
"Wang",
"Chengjie",
""
],
[
"Jin",
"Taisong",
""
],
[
"SevenShu",
"",
""
],
[
"Wu",
"Yunsheng",
""
],
[
"Liu",
"Yongge",
""
],
[
"Ji",
"Rongrong",
""
]
] |
2506.21102 | David Debot | David Debot, Pietro Barbiero, Gabriele Dominici, Giuseppe Marra | Interpretable Hierarchical Concept Reasoning through Attention-Guided Graph Learning | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Concept-Based Models (CBMs) are a class of deep learning models that provide interpretability by explaining predictions through high-level concepts. These models first predict concepts and then use them to perform a downstream task. However, current CBMs offer interpretability only for the final task prediction, while the concept predictions themselves are typically made via black-box neural networks. To address this limitation, we propose Hierarchical Concept Memory Reasoner (H-CMR), a new CBM that provides interpretability for both concept and task predictions. H-CMR models relationships between concepts using a learned directed acyclic graph, where edges represent logic rules that define concepts in terms of other concepts. During inference, H-CMR employs a neural attention mechanism to select a subset of these rules, which are then applied hierarchically to predict all concepts and the final task. Experimental results demonstrate that H-CMR matches state-of-the-art performance while enabling strong human interaction through concept and model interventions. The former can significantly improve accuracy at inference time, while the latter can enhance data efficiency during training when background knowledge is available. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 08:56:55 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Debot",
"David",
""
],
[
"Barbiero",
"Pietro",
""
],
[
"Dominici",
"Gabriele",
""
],
[
"Marra",
"Giuseppe",
""
]
] |
2506.21103 | Tim Lawson | Tim Lawson, Laurence Aitchison | Learning to Skip the Middle Layers of Transformers | 11 pages, 2 figures | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | Conditional computation is a popular strategy to make Transformers more efficient. Existing methods often target individual modules (e.g., mixture-of-experts layers) or skip layers independently of one another. However, interpretability research has demonstrated that the middle layers of Transformers exhibit greater redundancy, and that early layers aggregate information into token positions. Guided by these insights, we propose a novel architecture that dynamically skips a variable number of layers from the middle outward. In particular, a learned gating mechanism determines whether to bypass a symmetric span of central blocks based on the input, and a gated attention mechanism prevents subsequent tokens from attending to skipped token positions. Residual norms are controlled with a 'sandwich' or 'perilayernorm' scheme and gate sparsity with an adaptive regularization loss. We had aimed to reduce compute requirements for 'simpler' tokens and potentially foster an emergent multi-level representational hierarchy but, at the scales investigated, our approach does not achieve improvements in the trade-off between validation cross-entropy and estimated FLOPs compared to dense baselines with fewer layers. We release our code at https://github.com/tim-lawson/skip-middle. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:01:19 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Lawson",
"Tim",
""
],
[
"Aitchison",
"Laurence",
""
]
] |
2506.21104 | Wei Xie | Wei Xie, Viet Ha Hoang, Yin Yang, Yunqing Huang | Robust space-time multiscale upscaling via multicontinuum homogenization for evolving perforated media | null | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Time-evolving perforated domains arise in many engineering and geoscientific applications, including reactive transport, particle deposition, and structural degradation in porous media. Accurately capturing the macroscopic behavior of such systems poses significant computational challenges due to the dynamic fine-scale geometries. In this paper, we develop a robust and generalizable multiscale modeling framework based on multicontinuum homogenization to derive effective macroscopic equations in shrinking domains. The method distinguishes multiple continua according to the physical characteristics (e.g., channel widths), and couples them via space-time local cell problems formulated on representative volume elements. These local problems incorporate temporal derivatives and domain evolution, ensuring consistency with underlying fine-scale dynamics. The resulting upscaled system yields computable macroscopic coefficients and is suitable for large-scale simulations. Several numerical experiments are presented to validate the accuracy, efficiency, and potential applicability of the method to complex time-dependent engineering problems. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:01:53 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Xie",
"Wei",
""
],
[
"Hoang",
"Viet Ha",
""
],
[
"Yang",
"Yin",
""
],
[
"Huang",
"Yunqing",
""
]
] |
2506.21105 | Xiaoyu Xu | Xiaoyu Xu | On regularity of profinite isomorphisms between cusped hyperbolic 3-manifolds and the $A$-polynomial | 45 pages | null | null | null | math.GT math.GR | http://creativecommons.org/licenses/by/4.0/ | We prove that any isomorphism between the profinite completions of the fundamental groups of two cusped finite-volume hyperbolic 3-manifolds is regular and peripheral regular. As an application, we show that the $A$-polynomial of prime knots in $S^3$ is a profinite invariant, up to possible mirror image. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:02:52 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Xu",
"Xiaoyu",
""
]
] |
2506.21106 | Luis Felipe Casta\~no Ledesma | Felipe Casta\~no, Eduardo Fidalgo, Enrique Alegre, Rocio Alaiz-Rodr\'iguez, Raul Orduna and Francesco Zola | PhishKey: A Novel Centroid-Based Approach for Enhanced Phishing Detection Using Adaptive HTML Component Extraction | null | null | null | null | cs.CR cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Phishing attacks pose a significant cybersecurity threat, evolving rapidly to bypass detection mechanisms and exploit human vulnerabilities. This paper introduces PhishKey to address the challenges of adaptability, robustness, and efficiency. PhishKey is a novel phishing detection method using automatic feature extraction from hybrid sources. PhishKey combines character-level processing with Convolutional Neural Networks (CNN) for URL classification, and a Centroid-Based Key Component Phishing Extractor (CAPE) for HTML content at the word level. CAPE reduces noise and ensures complete sample processing avoiding crop operations on the input data. The predictions from both modules are integrated using a soft-voting ensemble to achieve more accurate and reliable classifications. Experimental evaluations on four state-of-the-art datasets demonstrate the effectiveness of PhishKey. It achieves up to 98.70% F1 Score and shows strong resistance to adversarial manipulations such as injection attacks with minimal performance degradation. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:04:55 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Castaño",
"Felipe",
""
],
[
"Fidalgo",
"Eduardo",
""
],
[
"Alegre",
"Enrique",
""
],
[
"Alaiz-Rodríguez",
"Rocio",
""
],
[
"Orduna",
"Raul",
""
],
[
"Zola",
"Francesco",
""
]
] |
2506.21107 | Chi Changxi | Changxi Chi and Jun Xia and Yufei Huang and Jingbo Zhou and Siyuan Li and Yunfan Liu and Chang Yu and Stan Z. Li | Unlasting: Unpaired Single-Cell Multi-Perturbation Estimation by Dual Conditional Diffusion Implicit Bridges | null | null | null | null | cs.LG q-bio.MN | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Estimating single-cell responses across various perturbations facilitates the identification of key genes and enhances drug screening, significantly boosting experimental efficiency. However, single-cell sequencing is a destructive process, making it impossible to capture the same cell's phenotype before and after perturbation. Consequently, data collected under perturbed and unperturbed conditions are inherently unpaired. Existing methods either attempt to forcibly pair unpaired data using random sampling, or neglect the inherent relationship between unperturbed and perturbed cells during the modeling. In this work, we propose a framework based on Dual Diffusion Implicit Bridges (DDIB) to learn the mapping between different data distributions, effectively addressing the challenge of unpaired data. We further interpret this framework as a form of data augmentation. We integrate gene regulatory network (GRN) information to propagate perturbation signals in a biologically meaningful way, and further incorporate a masking mechanism to predict silent genes, improving the quality of generated profiles. Moreover, gene expression under the same perturbation often varies significantly across cells, frequently exhibiting a bimodal distribution that reflects intrinsic heterogeneity. To capture this, we introduce a more suitable evaluation metric. We propose Unlasting, dual conditional diffusion models that overcome the problem of unpaired single-cell perturbation data and strengthen the model's insight into perturbations under the guidance of the GRN, with a dedicated mask model designed to improve generation quality by predicting silent genes. In addition, we introduce a biologically grounded evaluation metric that better reflects the inherent heterogeneity in single-cell responses. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:05:38 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Chi",
"Changxi",
""
],
[
"Xia",
"Jun",
""
],
[
"Huang",
"Yufei",
""
],
[
"Zhou",
"Jingbo",
""
],
[
"Li",
"Siyuan",
""
],
[
"Liu",
"Yunfan",
""
],
[
"Yu",
"Chang",
""
],
[
"Li",
"Stan Z.",
""
]
] |
2506.21108 | Lvzhou Li | Guanzhong Li, Jingquan Luo, Shiguang Feng, and Lvzhou Li | Deterministic quantum search on all Laplacian integral graphs | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Searching for an unknown marked vertex on a given graph (also known as spatial search) is an extensively discussed topic in the area of quantum algorithms, with a plethora of results based on different quantum walk models and targeting various types of graphs. Most of these algorithms have a non-zero probability of failure. In recent years, there have been some efforts to design quantum spatial search algorithms with $100\%$ success probability. However, these works either only work for very special graphs or only for the case where there is only one marked vertex. In this work, we propose a different and elegant approach to quantum spatial search, obtaining deterministic quantum search algorithms that can find a marked vertex with certainty on any Laplacian integral graph with any predetermined proportion of marked vertices. Thus, this work discovers the largest class of graphs so far that allow deterministic quantum search, making it easy to design deterministic quantum search algorithms for many graphs, including the different graphs discussed in previous works, in a unified framework. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:06:08 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Li",
"Guanzhong",
""
],
[
"Luo",
"Jingquan",
""
],
[
"Feng",
"Shiguang",
""
],
[
"Li",
"Lvzhou",
""
]
] |
2506.21109 | Luosheng Xu | Luosheng Xu, Dalin Zhang, Zhaohui Song | Pushing Trade-Off Boundaries: Compact yet Effective Remote Sensing Change Detection | 12 pages | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Remote sensing change detection is essential for monitoring urban expansion, disaster assessment, and resource management, offering timely, accurate, and large-scale insights into dynamic landscape transformations. While deep learning has revolutionized change detection, the increasing complexity and computational demands of modern models have not necessarily translated into significant accuracy gains. Instead of following this trend, this study explores a more efficient approach, focusing on lightweight models that maintain high accuracy while minimizing resource consumption, which is an essential requirement for on-satellite processing. To this end, we propose FlickCD, which means quick flick then get great results, pushing the boundaries of the performance-resource trade-off. FlickCD introduces an Enhanced Difference Module (EDM) to amplify critical feature differences between temporal phases while suppressing irrelevant variations such as lighting and weather changes, thereby reducing computational costs in the subsequent change decoder. Additionally, the FlickCD decoder incorporates Local-Global Fusion Blocks, leveraging Shifted Window Self-Attention (SWSA) and Enhanced Global Self-Attention (EGSA) to efficiently capture semantic information at multiple scales, preserving both coarse- and fine-grained changes. Extensive experiments on four benchmark datasets demonstrate that FlickCD reduces computational and storage overheads by more than an order of magnitude while achieving state-of-the-art (SOTA) performance or incurring only a minor (<1\% F1) accuracy trade-off. The implementation code is publicly available at https://github.com/xulsh8/FlickCD. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:06:52 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Xu",
"Luosheng",
""
],
[
"Zhang",
"Dalin",
""
],
[
"Song",
"Zhaohui",
""
]
] |
2506.21110 | Xun Zhou | Sheng Dai, Timo Kuosmanen, Xun Zhou | Orthogonality conditions for convex regression | null | null | null | null | stat.ME econ.EM | http://creativecommons.org/licenses/by/4.0/ | Econometric identification generally relies on orthogonality conditions, which usually state that the random error term is uncorrelated with the explanatory variables. In convex regression, the orthogonality conditions for identification are unknown. Applying Lagrangian duality theory, we establish the sample orthogonality conditions for convex regression, including additive and multiplicative formulations of the regression model, with and without monotonicity and homogeneity constraints. We then propose a hybrid instrumental variable control function approach to mitigate the impact of potential endogeneity in convex regression. The superiority of the proposed approach is shown in a Monte Carlo study and examined in an empirical application to Chilean manufacturing data. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:13:20 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Dai",
"Sheng",
""
],
[
"Kuosmanen",
"Timo",
""
],
[
"Zhou",
"Xun",
""
]
] |
2506.21111 | Chengchao Yuan | Chengchao Yuan, Leonard Pfeiffer, Walter Winter, Sara Buson, Federico Testagrossa, Jose Maria Sanchez Zaballa, Alessandra Azzollini | An Accretion Flare Interpretation for the UHE Neutrino Event KM3-230213A | 12 pages, 4 figures, 1 table | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | We study the origin of the ultra-high-energy (UHE) astrophysical neutrino event KM3-230213A detected by KM3NeT, focusing on MRC 0614-083 which has been pinpointed as the closest blazar to the neutrino localization. A joint interpretation of the optical, infrared, and X-ray light curves suggests that MRC 0614-083 has undergone a super-Eddington accretion flare accompanied by efficient proton acceleration. That flare has initiated a delayed infrared echo within the surrounding dust torus, which serves as a target for photomeson ($p\gamma$) interactions such that a self-consistent picture emerges: the predicted UHE neutrino flux is at the level expected from joint $E^{-2}$ fit with the IceCube measurements at lower energies, the variable nature of the event alleviates the tension with IceCube limits, and the accompanying electromagnetic cascade describes the X-ray flare around the neutrino detection time. Since a key remaining uncertainty is the unknown redshift of the source, we strongly encourage optical/ultraviolet spectroscopic measurements to determine its redshift. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:13:57 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Yuan",
"Chengchao",
""
],
[
"Pfeiffer",
"Leonard",
""
],
[
"Winter",
"Walter",
""
],
[
"Buson",
"Sara",
""
],
[
"Testagrossa",
"Federico",
""
],
[
"Zaballa",
"Jose Maria Sanchez",
""
],
[
"Azzollini",
"Alessandra",
""
]
] |
2506.21112 | Yancheng Wang | Yancheng Wang, Wei Guo, Guanying Chen, Ye Zhang, Shuguang Cui | Point Cloud Environment-Based Channel Knowledge Map Construction | null | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | Channel knowledge map (CKM) provides certain levels of channel state information (CSI) for an area of interest, serving as a critical enabler for environment-aware communications by reducing the overhead of frequent CSI acquisition. However, existing CKM construction schemes adopt over-simplified environment information, which significantly compromises their accuracy. To address this issue, this work proposes a joint model- and data-driven approach to construct CKM by leveraging point cloud environmental data along with a few samples of location-tagged channel information. First, we propose a novel point selector to identify subsets of point cloud that contain environmental information relevant to multipath channel gains, by constructing a set of co-focal ellipsoids based on different time of arrival (ToAs). Then, we trained a neural channel gain estimator to learn the mapping between each selected subset and its corresponding channel gain, using a real-world dataset we collected through field measurements, comprising environmental point clouds and corresponding channel data. Finally, experimental results demonstrate that: For CKM construction of power delay profile (PDP), the proposed method achieves a root mean squared error (RMSE) of 2.95 dB, significantly lower than the 7.32 dB achieved by the conventional ray-tracing method; for CKM construction of received power values, i.e., radio map, it achieves an RMSE of 1.04 dB, surpassing the Kriging interpolation method with an RMSE of 1.68 dB. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:15:59 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Wang",
"Yancheng",
""
],
[
"Guo",
"Wei",
""
],
[
"Chen",
"Guanying",
""
],
[
"Zhang",
"Ye",
""
],
[
"Cui",
"Shuguang",
""
]
] |
2506.21113 | Dominique Vuillaume | Joseane Santos Almeida, Sergio Gonz\'alez Casal, Hassan Al Sabea, Valentin Barth, Gautam Mitra, Vincent Delmas, David Gu\'erin, Olivier Galangau, Tiark Tiwary, Thierry Roisnel, Vincent Dorcet, Lucie Norel, Colin Van Dyck, Elke Scheer, Dominique Vuillaume, J\'er\^ome Cornil, St\'ephane Rigaut, Karine Costuas | Electronic and Thermoelectric Properties of Molecular Junctions Incorporating Organometallic Complexes: Implications for Thermoelectric Energy Conversion | null | null | null | null | cond-mat.mes-hall physics.app-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The electronic and thermoelectric properties of molecular junctions formed from iron and ruthenium metal-acetylide were studied using complementary experimental techniques and quantum chemical simulations. We performed physical characterizations of single-molecule and self-assembled monolayer junctions of the same molecules that allowed meaningful comparisons between the Ru and Fe adducts. In the case of the Fe-containing junctions, two distinct oxidation states are present. These junctions exhibit one of the highest Seebeck coefficients (S ca. 130 {\mu}V/K) reported to date for similar systems paired with broad electric conductance distribution and limited thermal conductance. As a result, the experimental thermoelectric figure of merit ZT for Fe-containing junctions reaches up to 0.4 for junctions with relatively high conductance. This is one of the highest ZT values reported for molecular systems at room temperature. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:19:10 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Almeida",
"Joseane Santos",
""
],
[
"Casal",
"Sergio González",
""
],
[
"Sabea",
"Hassan Al",
""
],
[
"Barth",
"Valentin",
""
],
[
"Mitra",
"Gautam",
""
],
[
"Delmas",
"Vincent",
""
],
[
"Guérin",
"David",
""
],
[
"Galangau",
"Olivier",
""
],
[
"Tiwary",
"Tiark",
""
],
[
"Roisnel",
"Thierry",
""
],
[
"Dorcet",
"Vincent",
""
],
[
"Norel",
"Lucie",
""
],
[
"Van Dyck",
"Colin",
""
],
[
"Scheer",
"Elke",
""
],
[
"Vuillaume",
"Dominique",
""
],
[
"Cornil",
"Jérôme",
""
],
[
"Rigaut",
"Stéphane",
""
],
[
"Costuas",
"Karine",
""
]
] |
2506.21114 | Mihai Prunescu | Mihai Prunescu | Polynomial fingerprinting for trees and formulas | null | null | null | null | math.LO | http://creativecommons.org/licenses/by/4.0/ | To cater to the needs of (Zero Knowledge) proofs for (mathematical) proofs, we describe a method to transform formal sentences in 2x2 - matrices over multivariate polynomials with integer coefficients, such that usual proof-steps like modus-ponens or the substitution are easy to compute from the matrices corresponding to the terms or formulas used as arguments. By evaluating the polynomial variables in random elements of a suitably chosen finite field, the proof is replaced by a numeric sequence. Only the values corresponding to the axioms have to be computed from scratch. The values corresponding to derived formulas are computed from the values corresponding to their ancestors by applying the homomorphic properties. On such sequences, various Zero Knowledge methods can be applied. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:20:21 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Prunescu",
"Mihai",
""
]
] |
2506.21115 | James Ramsey | James Ramsey, Faiza Alhamed, Jonathan P. Goss, Patrick R. Briddon, Mark J. Rayson | Reevaluating the electrical impact of atomic carbon impurities in MoS2 | 11 pages, 4 figures, 4 tables | null | null | null | cond-mat.mtrl-sci cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | Transition metal dichalcogenides, a family of two-dimensional compounds, are of interest for a range of technological applications. MoS2, the most researched member of this family, is hexagonal, from which monolayers may be isolated. Under ambient conditions and during growth/processing, contamination by impurities can occur, of which carbon is significant due to its presence in the common growth techniques. We have performed extensive computational investigations of carbon point defects, examining substitutional and interstitial locations. Previously unreported thermodynamically stable configurations, four-fold co-ordinated mono-carbon and di-carbon substitutions of Mo, and a complex of carbon substitution of sulfur bound to interstitial sulfur have been identified. We find no evidence to support recent assertions that carbon defects are responsible for electrical doping of MoS2, finding all energetically favorable forms have only deep charge transition levels and would act as carrier traps. To aid unambiguous identification of carbon defects, we present electronic and vibrational data for comparison with spectroscopy. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:23:15 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Ramsey",
"James",
""
],
[
"Alhamed",
"Faiza",
""
],
[
"Goss",
"Jonathan P.",
""
],
[
"Briddon",
"Patrick R.",
""
],
[
"Rayson",
"Mark J.",
""
]
] |
2506.21116 | Yujia Liang | Yujia Liang, Jile Jiao, Zhicheng Wang, Xuetao Feng, Zixuan Ye, Yuan Wang, Hao Lu | IPFormer-VideoLLM: Enhancing Multi-modal Video Understanding for Multi-shot Scenes | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Video Large Language Models (VideoLLMs) have demonstrated remarkable understanding capabilities, but are found struggling to tackle multi-shot scenarios,e.g., video clips with varying camera angles or scene changes. This challenge can render failures such as instance identity forgetting and key frame negligence. In this work, we first attribute the challenge to the lack of multi-shot annotations among existing datasets and therefore we introduce a new dataset termed MultiClip-Bench, featuring dense descriptions and instruction-based question-answering pairs tailored for multi-shot scenarios. We empirically find that the training set significantly boosts the multi-shot performance, while the testing benchmark provides a reliable measure of the model capability in multi-shot scenarios. By further analyzing and discovering that current models only encode instance features in a discrete or lossy manner, at the risk of missing identity information, we then contribute a new model IPFormer-VideoLLM. Its key idea is the injection of instance-level features as instance prompts through an efficient attention-based connector. This allows for the aggregation of instance-specific information across scenes. Experiments demonstrate that our proposed dataset and model not only enhance the multi-scene video understanding significantly, but also offer distinct advantages across various video benchmarks. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:30:57 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Liang",
"Yujia",
""
],
[
"Jiao",
"Jile",
""
],
[
"Wang",
"Zhicheng",
""
],
[
"Feng",
"Xuetao",
""
],
[
"Ye",
"Zixuan",
""
],
[
"Wang",
"Yuan",
""
],
[
"Lu",
"Hao",
""
]
] |
2506.21117 | Jan Ackermann | Jan Ackermann, Jonas Kulhanek, Shengqu Cai, Haofei Xu, Marc Pollefeys, Gordon Wetzstein, Leonidas Guibas, Songyou Peng | CL-Splats: Continual Learning of Gaussian Splatting with Local Optimization | ICCV 2025, Project Page: https://cl-splats.github.io | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In dynamic 3D environments, accurately updating scene representations over time is crucial for applications in robotics, mixed reality, and embodied AI. As scenes evolve, efficient methods to incorporate changes are needed to maintain up-to-date, high-quality reconstructions without the computational overhead of re-optimizing the entire scene. This paper introduces CL-Splats, which incrementally updates Gaussian splatting-based 3D representations from sparse scene captures. CL-Splats integrates a robust change-detection module that segments updated and static components within the scene, enabling focused, local optimization that avoids unnecessary re-computation. Moreover, CL-Splats supports storing and recovering previous scene states, facilitating temporal segmentation and new scene-analysis applications. Our extensive experiments demonstrate that CL-Splats achieves efficient updates with improved reconstruction quality over the state-of-the-art. This establishes a robust foundation for future real-time adaptation in 3D scene reconstruction tasks. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:32:37 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Ackermann",
"Jan",
""
],
[
"Kulhanek",
"Jonas",
""
],
[
"Cai",
"Shengqu",
""
],
[
"Xu",
"Haofei",
""
],
[
"Pollefeys",
"Marc",
""
],
[
"Wetzstein",
"Gordon",
""
],
[
"Guibas",
"Leonidas",
""
],
[
"Peng",
"Songyou",
""
]
] |
2506.21118 | Soh Kumabe | Tatsuya Gima, Soh Kumabe, and Yuichi Yoshida | Courcelle's Theorem for Lipschitz Continuity | ESA 2025, 27 pages | null | null | null | cs.DS | http://creativecommons.org/licenses/by/4.0/ | Lipschitz continuity of algorithms, introduced by Kumabe and Yoshida (FOCS'23), measures the stability of an algorithm against small input perturbations. Algorithms with small Lipschitz continuity are desirable, as they ensure reliable decision-making and reproducible scientific research. Several studies have proposed Lipschitz continuous algorithms for various combinatorial optimization problems, but these algorithms are problem-specific, requiring a separate design for each problem.
To address this issue, we provide the first algorithmic meta-theorem in the field of Lipschitz continuous algorithms. Our result can be seen as a Lipschitz continuous analogue of Courcelle's theorem, which offers Lipschitz continuous algorithms for problems on bounded-treewidth graphs. Specifically, we consider the problem of finding a vertex set in a graph that maximizes or minimizes the total weight, subject to constraints expressed in monadic second-order logic (MSO_2). We show that for any $\varepsilon>0$, there exists a $(1\pm \varepsilon)$-approximation algorithm for the problem with a polylogarithmic Lipschitz constant on bounded treewidth graphs. On such graphs, our result outperforms most existing Lipschitz continuous algorithms in terms of approximability and/or Lipschitz continuity. Further, we provide similar results for problems on bounded-clique-width graphs subject to constraints expressed in MSO_1. Additionally, we construct a Lipschitz continuous version of Baker's decomposition using our meta-theorem as a subroutine. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:34:09 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Gima",
"Tatsuya",
""
],
[
"Kumabe",
"Soh",
""
],
[
"Yoshida",
"Yuichi",
""
]
] |
2506.21119 | Xiaoshuang Ji | Xiaoshuang Ji, Zhendong Zhao, Xiaojun Chen, Xin Zhao, and Zeyao Liu | Progtuning: Progressive Fine-tuning Framework for Transformer-based Language Models | Accepted by ICONIP 2024 | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Fine-tuning is a promising technique for leveraging Transformer-based language models in downstream tasks. As model sizes continue to grow, updating all model parameters becomes increasingly costly. Parameter-efficient fine-tuning methods effectively address this issue by selectively updating a small subset of parameters. However, fine-tuning and most existing parameter-efficient fine-tuning methods require updating the same number of parameters as the initial size, ignoring the unequal contribution across Transformer blocks and leading to extremely inefficient allocation of computing resources. In this paper, we propose Progtuning, the novel fine-tuning framework combined with progressive learning for Transformer-based language models. Specifically, Progtuning progressively reduces the number of updated transformer blocks based on the contribution. Remarkably, Progtuning optimizes resource allocation and reduces the number of updated parameters by approximately 25\%, while still maintaining competitive performance. And it also exhibits high adaptability with parameter-efficient fine-tuning methods, demonstrating excellent performance across various adaptation scenarios. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:37:15 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Ji",
"Xiaoshuang",
""
],
[
"Zhao",
"Zhendong",
""
],
[
"Chen",
"Xiaojun",
""
],
[
"Zhao",
"Xin",
""
],
[
"Liu",
"Zeyao",
""
]
] |
2506.21120 | Christopher Lustri | Christopher J. Lustri, In\^es Aniceto, Panayotis G. Kevrekidis | Borel-Pad\'e exponential asymptotics for the discrete nonlinear Schr\"odinger model with next-to-nearest neighbour interactions | 19 pages, 5 figures | null | null | null | nlin.PS math-ph math.MP nlin.SI | http://creativecommons.org/licenses/by/4.0/ | In the present work we study discrete nonlinear Schr{\"o}dinger models combining nearest (NN) and next-nearest (NNN) neighbor interactions, motivated by experiments in waveguide arrays. While we consider the more experimentally accessible case of positive ratio $\mu$ of NNN to NN interactions, we focus on the intriguing case of competing such interactions $(\mu<0)$, where stationary states can exist only for $-1/4 < \mu < 0$. We analyze the key eigenvalues for the stability of the pulse-like stationary (ground) states, and find that such modes depend exponentially on the coupling parameter $\eps$, with suitable polynomial prefactors and corrections that we analyze in detail. Very good agreement of the resulting predictions is found with systematic numerical computations of the associated eigenvalues. This analysis uses Borel-Pad\'{e} exponential asymptotics to determine Stokes multipliers in the solution; these multipliers cannot be obtained using standard matched asymptotic expansion approaches as they are hidden beyond all asymptotic orders, even near singular points. By using Borel-Pad\'{e} methods near the singularity, we construct a general asymptotic template for studying parametric problems which require the calculation of subdominant Stokes multipliers. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:41:00 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Lustri",
"Christopher J.",
""
],
[
"Aniceto",
"Inês",
""
],
[
"Kevrekidis",
"Panayotis G.",
""
]
] |
2506.21121 | Muleilan Pei | Muleilan Pei, Shaoshuai Shi, Lu Zhang, Peiliang Li, Shaojie Shen | GoIRL: Graph-Oriented Inverse Reinforcement Learning for Multimodal Trajectory Prediction | Accepted by ICML 2025 | null | null | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Trajectory prediction for surrounding agents is a challenging task in autonomous driving due to its inherent uncertainty and underlying multimodality. Unlike prevailing data-driven methods that primarily rely on supervised learning, in this paper, we introduce a novel Graph-oriented Inverse Reinforcement Learning (GoIRL) framework, which is an IRL-based predictor equipped with vectorized context representations. We develop a feature adaptor to effectively aggregate lane-graph features into grid space, enabling seamless integration with the maximum entropy IRL paradigm to infer the reward distribution and obtain the policy that can be sampled to induce multiple plausible plans. Furthermore, conditioned on the sampled plans, we implement a hierarchical parameterized trajectory generator with a refinement module to enhance prediction accuracy and a probability fusion strategy to boost prediction confidence. Extensive experimental results showcase our approach not only achieves state-of-the-art performance on the large-scale Argoverse & nuScenes motion forecasting benchmarks but also exhibits superior generalization abilities compared to existing supervised models. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:46:53 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Pei",
"Muleilan",
""
],
[
"Shi",
"Shaoshuai",
""
],
[
"Zhang",
"Lu",
""
],
[
"Li",
"Peiliang",
""
],
[
"Shen",
"Shaojie",
""
]
] |
2506.21122 | Nicetu Tibau Vidal | Nicetu Tibau Vidal, Chiara Marletto, Vlatko Vedral, GIulio Chiribella | Bose-Marletto-Vedral experiment without observable spacetime superpositions | 12+2 pages, 5 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Reconciling quantum mechanics and general relativity remains one of the most profound challenges in modern physics. The BMV (Bose-Marletto-Vedral) experiment can assess the quantum nature of gravity by testing whether gravitational interactions can generate entanglement between quantum systems. In this work, we show that entanglement can be generated by gravity without requiring spacetime superpositions or quantum spacetime degrees of freedom by using mediators that do not satisfy the usual property of local tomography when coupling to quantum matter. Specifically, we showcase how entanglement can be generated using three distinct toy models that display non-locally tomographic couplings between quantum matter and a locally classical gravitational mediator. These models include (i) fermionic systems with the parity superselection rule, (ii) non-Abelian anyonic systems, and (iii) a novel bit anti-bit model. Our results demonstrate a crucial point: a gravitational mediator which does not exhibit superpositions of its classical basis but still qualifies as non-classical via non-locally tomographic coupling mechanisms can generate entanglement through local interactions. This work also underscores the importance of relaxing local tomography in exploring the quantum-gravitational interface. It provides a novel perspective on the role of spacetime degrees of freedom in entanglement generation through local interactions. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:48:55 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Vidal",
"Nicetu Tibau",
""
],
[
"Marletto",
"Chiara",
""
],
[
"Vedral",
"Vlatko",
""
],
[
"Chiribella",
"GIulio",
""
]
] |
2506.21123 | Hao Wu | Hao Wu, Chongwu Xie, Xinyuan Yao, Kang-Da Wu, Shanchi Wu, Rui Ni, Guo-Yong Xiang, Chen Gong | Characterization of Rydberg-Atom Signal Reception of Dual-Frequency Signals Coupled with Two Energy Levels | null | null | null | null | eess.SP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Rydberg atomic sensors have been adopted for novel radio frequency (RF) measurement technique and the sensing capability for signals in multiple frequencies makes it attractive for multi-user communication. However, unlike traditional antennas where the signals in multiple frequencies are orthogonal, the received signals of atomic sensors corresponding to different energy levels will be downconverted to the baseband simultaneously, resulting in multi-user interference. Thus, in this paper, we analyze the mutual interference characteristics of two RF signals with different carrier frequencies coupling different energy levels. We introduce the joint response coefficient based on the receiver characteristics and analyze the interference of one user to another. We analyze the bit-error rate (BER) and symbol-error rate (SER) for two signals coupling two different energy levels. We also conduct experiments to validate the BER and SER results. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:50:00 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Wu",
"Hao",
""
],
[
"Xie",
"Chongwu",
""
],
[
"Yao",
"Xinyuan",
""
],
[
"Wu",
"Kang-Da",
""
],
[
"Wu",
"Shanchi",
""
],
[
"Ni",
"Rui",
""
],
[
"Xiang",
"Guo-Yong",
""
],
[
"Gong",
"Chen",
""
]
] |
2506.21124 | Salvatore Cuomo | G. Intoccia, U. Chirico, V. Schiano Di Cola, G. Pepe, S. Cuomo | Quantum Adaptive Search: A Hybrid Quantum-Classical Algorithm for Global Optimization of Multivariate Functions | null | null | null | null | quant-ph cs.NA math.NA math.OC | http://creativecommons.org/licenses/by/4.0/ | This work presents Quantum Adaptive Search (QAGS), a hybrid quantum-classical algorithm for the global optimization of multivariate functions. The method employs an adaptive mechanism that dynamically narrows the search space based on a quantum-estimated probability distribution of the objective function. A quantum state encodes information about solution quality through an appropriate complex amplitude mapping, enabling the identification of the most promising regions, and thus progressively tightening the search bounds; then a classical optimizer performs local refinement of the solution. The analysis demonstrates that QAGS ensures a contraction of the search space toward global optima, with controlled computational complexity. The numerical results on the benchmark functions show that, compared to the classical methods, QAGS achieves higher accuracy while offering advantages in both time and space complexity. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:55:36 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Intoccia",
"G.",
""
],
[
"Chirico",
"U.",
""
],
[
"Di Cola",
"V. Schiano",
""
],
[
"Pepe",
"G.",
""
],
[
"Cuomo",
"S.",
""
]
] |
2506.21125 | Pankaj Mishra | Swarup K. Sarkar, Sh. Mardonov, E. Ya. Sherman, and Pankaj K. Mishra | Rabi-induced localization and resonant delocalization of a binary condensate in a spin-asymmetric quasiperiodic potential | 15pages, 15figures | null | null | null | cond-mat.quant-gas | http://creativecommons.org/licenses/by/4.0/ | We theoretically investigate the ground state and dynamics of a Rabi-coupled pseudospin-1/2 Bose-Einstein condensate, where only one spin component is subjected to an external potential. We show that in the quasiperiodic potential the Rabi coupling induces localization between the components as it is raised above the threshold value. Interestingly, the localization is mutually induced by both components for the quasiperiodic confinement, whereas for a harmonic trap the localization is induced in the potential-free component by interaction with that confined in the potential. Further, we explore the condensate dynamics by implementing a periodic driving of the Rabi frequency, where various frequency-dependent delocalization patterns, such as double (triple)-minima, tree-(parquet)-like, and frozen distributions with a correlated propagation of different spin populations are observed in the condensate density. These features pave the way to control the condensate mass and spin density patterns, both in the stationary and dynamical realizations. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:59:23 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Sarkar",
"Swarup K.",
""
],
[
"Mardonov",
"Sh.",
""
],
[
"Sherman",
"E. Ya.",
""
],
[
"Mishra",
"Pankaj K.",
""
]
] |
2506.21126 | Jiajia Guo | Jiajia Guo, Yiming Cui, Shi Jin | Semantic-aware Digital Twin for AI-based CSI Acquisition | This article has been accepted by IEEE Communications Standards Magazine | null | null | null | cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Artificial intelligence (AI) substantially enhances channel state information (CSI) acquisition performance but is limited by its reliance on single-modality information and deployment challenges, particularly in dataset collection. This paper investigates the use of semantic-aware digital twin (DT) to enhance AI-based CSI acquisition. We first briefly introduce the motivation and recent advancements in AI-driven CSI acquisition and semantic-aware DT employment for air interfaces. Then, we thoroughly explore how semantic-aware DT can bolster AI-based CSI acquisition. We categorizes the semantic-aware DT for AI-based CSI acquisition into two classes: enhancing AI-based CSI acquisition through integration with DT and using DT to aid AI-based CSI deployment. Potential integration frameworks are introduced in detail. Finally, we conclude by outlining potential research directions within the semantic-aware DT-assisted AI-based CSI acquisition. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 09:59:26 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Guo",
"Jiajia",
""
],
[
"Cui",
"Yiming",
""
],
[
"Jin",
"Shi",
""
]
] |
2506.21127 | Deepak Kumar Panda | Deepak Kumar Panda and Weisi Guo | Robust Policy Switching for Antifragile Reinforcement Learning for UAV Deconfliction in Adversarial Environments | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | The increasing automation of navigation for unmanned aerial vehicles (UAVs) has exposed them to adversarial attacks that exploit vulnerabilities in reinforcement learning (RL) through sensor manipulation. Although existing robust RL methods aim to mitigate such threats, their effectiveness has limited generalization to out-of-distribution shifts from the optimal value distribution, as they are primarily designed to handle fixed perturbation. To address this limitation, this paper introduces an antifragile RL framework that enhances adaptability to broader distributional shifts by incorporating a switching mechanism based on discounted Thompson sampling (DTS). This mechanism dynamically selects among multiple robust policies to minimize adversarially induced state-action-value distribution shifts. The proposed approach first derives a diverse ensemble of action robust policies by accounting for a range of perturbations in the policy space. These policies are then modeled as a multiarmed bandit (MAB) problem, where DTS optimally selects policies in response to nonstationary Bernoulli rewards, effectively adapting to evolving adversarial strategies. Theoretical framework has also been provided where by optimizing the DTS to minimize the overall regrets due to distributional shift, results in effective adaptation against unseen adversarial attacks thus inducing antifragility. Extensive numerical simulations validate the effectiveness of the proposed framework in complex navigation environments with multiple dynamic three-dimensional obstacles and with stronger projected gradient descent (PGD) and spoofing attacks. Compared to conventional robust, non-adaptive RL methods, the antifragile approach achieves superior performance, demonstrating shorter navigation path lengths and a higher rate of conflict-free navigation trajectories compared to existing robust RL techniques | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 10:06:29 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Panda",
"Deepak Kumar",
""
],
[
"Guo",
"Weisi",
""
]
] |
2506.21128 | Sara Kalisnik Hintz | Sara Kali\v{s}nik and Davorin Le\v{s}nik | Tractable Metric Spaces and the Continuity of Magnitude | null | null | null | null | math.GN | http://creativecommons.org/licenses/by/4.0/ | Magnitude is an isometric invariant of metric spaces introduced by Leinster. Since its inception, it has inspired active research into its connections with integral geometry, geometric measure theory, fractal dimensions, persistent homology, and applications in machine learning. In particular, when it comes to applications, continuity and stability of invariants play an important role. Although it has been shown that magnitude is nowhere continuous on the Gromov--Hausdorff space of finite metric spaces, positive results are possible if we restrict the ambient space. In this paper, we introduce the notion of tractable metric spaces, provide a characterization of these spaces, and establish several continuity results for magnitude in this setting. As a consequence, we offer a new proof of a known result stating that magnitude is continuous on the space of compact subsets of $\mathbb{R}$ with respect to the Hausdorff metric. Furthermore, we show that the magnitude function is Lipschitz when restricted to bounded subspaces of $\mathbb{R}$. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 10:08:47 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Kališnik",
"Sara",
""
],
[
"Lešnik",
"Davorin",
""
]
] |
2506.21129 | Deepak Kumar Panda | Deepak Kumar Panda, Adolfo Perrusquia and Weisi Guo | Curriculum-Guided Antifragile Reinforcement Learning for Secure UAV Deconfliction under Observation-Space Attacks | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reinforcement learning (RL) policies deployed in safety-critical systems, such as unmanned aerial vehicle (UAV) navigation in dynamic airspace, are vulnerable to out-ofdistribution (OOD) adversarial attacks in the observation space. These attacks induce distributional shifts that significantly degrade value estimation, leading to unsafe or suboptimal decision making rendering the existing policy fragile. To address this vulnerability, we propose an antifragile RL framework designed to adapt against curriculum of incremental adversarial perturbations. The framework introduces a simulated attacker which incrementally increases the strength of observation-space perturbations which enables the RL agent to adapt and generalize across a wider range of OOD observations and anticipate previously unseen attacks. We begin with a theoretical characterization of fragility, formally defining catastrophic forgetting as a monotonic divergence in value function distributions with increasing perturbation strength. Building on this, we define antifragility as the boundedness of such value shifts and derive adaptation conditions under which forgetting is stabilized. Our method enforces these bounds through iterative expert-guided critic alignment using Wasserstein distance minimization across incrementally perturbed observations. We empirically evaluate the approach in a UAV deconfliction scenario involving dynamic 3D obstacles. Results show that the antifragile policy consistently outperforms standard and robust RL baselines when subjected to both projected gradient descent (PGD) and GPS spoofing attacks, achieving up to 15% higher cumulative reward and over 30% fewer conflict events. These findings demonstrate the practical and theoretical viability of antifragile reinforcement learning for secure and resilient decision-making in environments with evolving threat scenarios. | [
{
"version": "v1",
"created": "Thu, 26 Jun 2025 10:10:41 GMT"
}
] | 2025-06-27T00:00:00 | [
[
"Panda",
"Deepak Kumar",
""
],
[
"Perrusquia",
"Adolfo",
""
],
[
"Guo",
"Weisi",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.