中 EN
登录 注册账户

Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达，利用大语言模型自动生成交叉领域文献分析简报。

登录 注册账户
01.
arXiv (math.PR) 2026-06-15

Boltzmann-Like Occupation of Nonequilibrium Steady States on Dense Networks

作者:
Jacob Calvert↗

arXiv:2606.14542v1 Announce Type: cross Abstract: A central problem in statistical physics is to extend the Boltzmann distribution to nonequilibrium steady states (NESS). We prove that NESS on large dense networks have Boltzmann-like occupation despite extensive entropy production. We further show that the active-matter heuristic of "low rattling" is asymptotically exact. Intuitively, these NESS spend a greater fraction of their time in states they leave more slowly. This explanation extends to the broader class of "equiaccessible" steady states, which play a role in our analysis akin to that of equilibrium in linear response.

阅读与讨论 → 访问原文 →
02.
arXiv (CS.AI) 2026-06-19

Which Pairs to Compare for LLM Post-Training?

作者:
Jiangze Han↗Vineet Goyal↗Will Ma↗

arXiv:2606.19607v1 Announce Type: new Abstract: Preference-based post-training has become a central paradigm for aligning language models. A common data-collection strategy is to generate a small set of completions for each prompt and label the resulting comparison pairs. However, human preference labels are often much more expensive than generating additional completions, suggesting a different use of the same labeling budget: generate a larger pool of completions, but label only the most informative comparison pairs. This paper studies which pairs should be compared in preference-based post-training. We formulate comparison curation as a sampling-design problem and evaluate designs by the quality of the final policy under the preference-based post-training objective. We instantiate this framework for Direct Preference Optimization (DPO), analyzing how the choice of labeled pairs propagates through DPO training to downstream policy performance. Our main results provide matching upper and lower bounds on the post-training optimality gap of the DPO-trained policy. The bounds show that comparison selection affects downstream performance through a single design-dependent information matrix, which links label allocation to parameter estimation error and policy suboptimality. This yields an explicit optimization criterion for budgeted comparison curation and motivates practical sampling designs for selecting informative pairs from large generated completion pools. Experiments on synthetic settings and language-model post-training benchmarks show that the proposed designs consistently improve sample efficiency over common comparison-selection heuristics.

阅读与讨论 → 访问原文 →
04.
arXiv (CS.CL) 2026-06-19

S2D2: Fast Decoding for Diffusion LLMs via Training-Free Self-Speculation

作者:
Ligong Han↗Hao Wang↗Han Gao↗Kai Xu↗Akash Srivastava↗

Block-diffusion language models offer a promising path toward faster-than-autoregressive generation by combining block-wise autoregressive decoding with within-block parallel denoising. However, in the few-step regime needed for practical acceleration, standard confidence-thresholded decoding is often brittle: aggressive thresholds hurt quality, while conservative thresholds require unnecessary denoising steps. Existing approaches that address this issue either require additional training or incur extra test-time compute. We present S2D2, a training-free self-speculative decoding framework for block-diffusion language models. Our key observation is that a block-diffusion model becomes autoregressive when the block size is reduced to one, allowing the same pretrained model to act as both drafter and verifier. S2D2 inserts a speculative verification step into standard block-diffusion decoding and uses lightweight routing policies to decide when verification is worth its cost. This yields a hybrid decoding trajectory in which diffusion proposes tokens in parallel, while the autoregressive mode acts as a local sequence-level critic. Across three mainstream block-diffusion families, S2D2 consistently improves the accuracy-speed tradeoff over strong confidence-thresholding baselines. On SDAR, we observe up to $4.7\times$ speedup over autoregressive decoding, and up to $1.57\times$ over a tuned dynamic decoding baseline while improving accuracy by up to $4.5$ points. On LLaDA2.1-Mini, S2D2 remains complementary to built-in self-correction, including a conservative setting where it is $4.4\times$ faster than the static baseline with slightly higher accuracy.

阅读与讨论 → 访问原文 →
05.
arXiv (CS.CV) 2026-06-11

ReMoT: Reinforcement Learning with Motion Contrast Triplets

作者:
Cong Wan↗Zeyu Guo↗Jiangyang Li↗SongLin Dong↗Yifan Bai↗Lin Peng↗Zhiheng Ma↗Yihong Gong↗

We present ReMoT, a unified training paradigm to systematically address the fundamental shortcomings of VLMs in spatio-temporal consistency – a critical failure point in navigation, robotics, and autonomous driving. ReMoT integrates two core components: (1) A rule-based automatic framework that generates ReMoT-16K, a large-scale (16.5K triplets) motion-contrast dataset derived from video meta-annotations, surpassing costly manual or model-based generation. (2) Group Relative Policy Optimization, which we empirically validate yields optimal performance and data efficiency for learning this contrastive reasoning, far exceeding standard Supervised Fine-Tuning. We also construct the first benchmark for fine-grained motion contrast triplets to measure a VLM's discrimination of subtle motion attributes (e.g., opposing directions). The resulting model achieves state-of-the-art performance on our new benchmark and multiple standard VLM benchmarks, culminating in a remarkable 25.1% performance leap on spatio-temporal reasoning tasks.

阅读与讨论 → 访问原文 →
06.
arXiv (CS.CV) 2026-06-16

AURA: Active-Response Attribution under Treatment Ambiguity in Bacterial Cytological Profiling

作者:
Kartik Jhawar↗Mrunmayee Deshpande↗Wilfried Moreira↗Guillermo C. Bazan↗Lipo Wang↗

When a bacterial sample is exposed to several antibiotics, not every applied drug necessarily acts: if the organism is resistant to one of them, that drug leaves no morphological trace. The clinically meaningful quantity is therefore not which antibiotics were applied, but which ones were active. We show that these two are sharply decoupled in real E. coli microscopy - naively assuming the applied combination equals the active one is correct only about 37% of the time - yet existing computational tools are ill-suited to recovering the active set. Forward perturbation models such as scGen, CPA, and IMPA are designed to predict appearance from treatment, not the reverse, and inverting them degrades sharply; discriminative image classifiers tend to memorise strain- and batch-specific texture and fail to transfer across experimental replicates. We introduce AURA, which reframes the task as constrained, energy-based inverse attribution. Its central inductive bias is that the active set must be a subset of the applied set; this collapses the candidate space and lets AURA infer the active subset of applied antibiotics by decomposing residual morphology into antibiotic response atoms and selecting the subset with the lowest reconstruction energy, using no strain label at test time. AURA-E adds evidence-aware abstention, withholding a prediction when candidate explanations remain near-equally plausible. On cross-replicate transfer in an E. coli cytological profiling dataset, AURA recovers the active antibiotic combination with 95.47% exact-match accuracy.

阅读与讨论 → 访问原文 →
07.
arXiv (CS.CL) 2026-06-12

M\"OVE: A Holistic LLM Benchmark for the German Public Sector

作者:
Camilla Dalerci↗Thilo Michael↗Robin Schaefer↗Daniel Weinland↗

We present M\"OVE (Modelle für die \"Offentliche Verwaltung Evaluieren), a holistic benchmark for evaluating large language models (LLMs) in the context of the German public sector. While LLMs are increasingly adopted in public administration, model selection remains largely ad hoc, and existing benchmarks offer limited guidance: they are predominantly English-centric, US-centric in content, and focus exclusively on task performance. M\"OVE addresses these gaps by evaluating 39 models across two complementary dimensions. Performance criteria cover summarization, question answering, and topic extraction. Governance criteria assess hallucination tendencies, energy consumption, provider transparency, and alignment with German constitutional values and knowledge about positions by German political parties. In total, we utilize ten German-language datasets, including gold- and silverstandard datasets that we constructed to reflect public-administration domains. We employ a multi-metric evaluation strategy combining classical NLP metrics, embedding-based methods, and LLM-as-a-judge approaches. Our results show that no single model dominates across all criteria: top performers differ between tasks, and model size alone is a poor predictor of quality. We further evaluate the benchmark itself, analyzing its statistical precision, LLM judge reliability, the impact of our private datasets on model rankings, the sensitivity of our results to prompt formulation, and the validity of our energy consumption estimates. M\"OVE is designed as a living benchmark under active development; results are publicly available at https://moeve.bundesdruckerei.de/.

阅读与讨论 → 访问原文 →
08.
arXiv (CS.CV) 2026-06-15

Overhead Wildlife Locator (OWL): Benchmarking Weakly Supervised Learning for Aerial Wildlife Surveys

作者:
Isai Daniel Chac\'on↗Zhongqi Miao↗Bruno Demuro↗Caleb Robinson↗Rahul Dodhia↗Lasha Otarashvili↗Jason Holmberg↗Kirk Larsen↗Howard Frederick↗Nathan J. Pamperin↗Pablo Arbel\'aez↗Juan M. Lavista Ferres↗…

Automated aerial wildlife surveys increasingly rely on deep learning, yet standard object detectors require bounding-box annotations, reported to be up to seven times slower and three times more expensive to produce than point-level labels. To address this bottleneck, we introduce the Overhead Wildlife Locator (OWL), a weakly supervised density-estimation framework with three variants: OWL-C, a fully convolutional model for high-throughput screening; OWL-T, a Swin-augmented hybrid for heterogeneous, cluttered scenes; and OWL-D, built on a frozen DINOv3 ViT-H+/16 encoder with a DPT-style fusion decoder. We benchmark all three against POLO, YOLOv11n, and YOLOv11l across five public aerial datasets, from sparse fixed-wing savanna surveys to dense UAV paddock imagery, and against the published HerdNet baseline on its native Delplanque split. OWL-D sets a new state of the art on Delplanque (0.934 AP vs. HerdNet's 0.840) and records the highest AP on four of the five datasets. Performance is regime-dependent: on the extreme-density SheepCounter UAV dataset the hybrid OWL-T leads (0.978 AP) and the convolutional variants attain the lowest counting error, whereas the foundation-based OWL-D degrades, indicating which variant suits which survey type. We further validate operational readiness on the Alaska Department of Fish and Game's 2022 Central Arctic Caribou census: under cross-herd and cross-temporal transfer, OWL-C fine-tuned on the 2017 Porcupine Caribou Herd split attains F1 = 0.965 on a held-out patch test set, with a signed count error of +3.1% aggregated across the released test patches. We release the OWL code, model weights, and the annotated Porcupine Caribou Herd 2017 (PCH) and Central Arctic Herd 2022 (CAH) patches, the first open patch-level datasets for large-scale caribou aerial surveys, at https://github.com/microsoft/MegaDetector-Overhead.

阅读与讨论 → 访问原文 →
09.
arXiv (CS.AI) 2026-06-11

Generalizing Beyond Suboptimality: Offline Reinforcement Learning Learns Effective Scheduling through Random Solutions

作者:
Jesse van Remmerden↗Zaharah Bukhsh↗Yingqian Zhang↗

arXiv:2509.10303v2 Announce Type: replace-cross Abstract: Online reinforcement learning (RL) approaches have demonstrated strong performance on Job Shop Scheduling (JSP) and Flexible JSP (FJSP) problems by learning scheduling policies through direct interaction with simulated environments. However, these methods often require extensive training interactions, limiting their sample efficiency and practical applicability. Motivated by this challenge, we introduce Conservative Discrete Quantile Actor-Critic (CDQAC), an offline RL algorithm that learns effective scheduling policies directly from static, suboptimal datasets. CDQAC couples a quantile-based critic with delayed policy updates to estimate the return distribution of machine-operation pairs. Extensive experiments on JSP and FJSP benchmarks demonstrate that CDQAC consistently outperforms the data-generating heuristics, surpasses state-of-the-art offline and online RL baselines, and is highly sample efficient, requiring only 1 to 5% of the original dataset to learn high-quality policies. Our analysis suggests that, in scheduling, offline RL performance is governed mainly by state-action coverage rather than the quality of individual trajectories. Scheduling couples a dense reward aligned with the makespan objective with equal-length trajectories across heuristics, enabling effective learning from a broad range of behaviors. Consistent with this observation, datasets generated by a simple random heuristic with broader coverage let it outperform policies trained on datasets produced by stronger heuristics such as Genetic Algorithms.

阅读与讨论 → 访问原文 →
10.
arXiv (CS.AI) 2026-06-18

From Memorization to Parameter Interference: How Overtraining Experts Harms Model Merging

作者:
Stefan Horoi↗Guy Wolf↗Eugene Belilovsky↗Gintare Karolina Dziugaite↗

arXiv:2506.14126v2 Announce Type: replace-cross Abstract: Modern deep learning is increasingly characterized by the use of open-weight foundation models that can be fine-tuned on specialized datasets. This has led to a proliferation of expert models and adapters, often shared via platforms like HuggingFace and AdapterHub. Model merging has recently emerged as an effective way to leverage these existing resources, enabling the composition of capabilities from different model checkpoints. A natural pipeline has thus formed to harness the benefits of transfer learning and amortize sunk training costs: models are pre-trained on general data, fine-tuned on specific tasks, and then multiple checkpoints are merged to obtain a more capable model. A prevailing assumption is that improvements at one stage of this pipeline propagate downstream, leading to gains at subsequent steps. In this work, we challenge that assumption by examining how expert fine-tuning affects model merging. We show that long fine-tuning of experts that optimizes for their individual performance leads to degraded merging performance across vision and language modalities, multiple model scales, and both fully fine-tuned and LoRA-adapted models. We trace this degradation to the memorization of a small set of difficult examples that dominate late fine-tuning steps. This causes negative parameter interference and encodes knowledge that is forgotten during merging. Finally, we demonstrate that task-dependent aggressive early stopping strategies can significantly improve model merging performance.

阅读与讨论 → 访问原文 →
11.
arXiv (CS.LG) 2026-06-15

Time Series Causal Discovery via Context-Conditioned and Causality-Augmented Pretraining

作者:
Biao Ouyang↗Tengxue Zhang↗Zhihao Zhuang↗Yang Shu↗Chenjuan Guo↗Bin Yang↗

arXiv:2605.26759v2 Announce Type: replace Abstract: Causal discovery from time series is critical for many real-world applications, such as tracing the root causes of anomalies. Existing approaches typically rely on dataset-specific optimization, making it difficult to transfer their causal discovery capabilities to new time series governed by diverse causal mechanisms. In this paper, we propose PTCD, a novel Pretraining framework for Time-series Causal Discovery, which improves cross-task generalization through context-conditioned modeling and transferable causal augmentation. To model complex temporal causal dependencies, PTCD employs a dual-scale iterative attention mechanism to capture window-level causal relationships, and a Gaussian mixture with a context-level routing mechanism to handle heterogeneous exogenous distributions. To further address distribution shifts across causal graphs, PTCD adopts a pretraining paradigm on synthetic datasets that integrates intervention-based learning and a causal mixup strategy, promoting stable causal discovery and stronger generalization. Extensive experiments on multiple real-world out-of-distribution (OOD) datasets demonstrate that PTCD excels in both causal discovery and root cause identification.

阅读与讨论 → 访问原文 →
12.
medRxiv (Medicine) 2026-06-16

Development of a symptom-based severity score anchored to health-related quality of life post-COVID-19 within the population-based EPILOC cohorts

作者:
Peter↗R. S↗Sedelmaier↗Nieters↗Schilling↗Matits↗Goepel↗Merle↗Steinacker↗J. M↗Kern↗W. V↗…

Purpose Because simple symptom counts treat all symptoms as equally important and may not adequately capture the HRQoL impact of heterogeneous post-COVID-19 symptoms, we aimed to develop an HRQoL-anchored symptom severity score providing an interpretable measure of post-COVID-19 disease burden. Methods Baseline data from the population-based EPILOC and EPILOC Omicron surveys (adults aged 18-65 years) were used to develop a symptom-based severity score anchored to physical and mental HRQoL assessed with the SF-12. A two-stage modelling approach was applied to identify HRQoL-relevant symptoms and to derive symptom-specific weights for physical and mental component scores, incorporating 30 ordinal symptom severity variables. Symptom-specific weights were extracted to compute physical, mental, and composite severity scores. Score interpretation was examined using external reference measures, including EPILOC case status, self-reported health recovery, and functional consequences. Results A total of 19,004 participants (mean age 44.3 years, 59.6% female) were included. Sixteen symptoms contributed to the physical and eleven to the mental HRQoL score, with a limited subset accounting for most of the HRQoL loss. Severity scores were heavily right-skewed, with 50.6% of participants showing no measurable HRQoL impairment. Higher scores correlated with lower self-reported recovery, and increased probability of rehabilitation use and health-related changes in working time, supporting convergent and criterion-related validity. Conclusions This study introduces a transparent, HRQoL-anchored symptom severity score that measures graded post-COVID-19 burden beyond simple symptom counts. The score may be particularly suited for longitudinal assessment of recovery trajectories.

阅读与讨论 → 访问原文 →
13.
arXiv (CS.LG) 2026-06-19

Comparing Linear Probes with Mahalanobis Cosine Similarity

作者:
Zhuofan Josh Ying↗Peter Hase↗Nikolaus Kriegeskorte↗

arXiv:2606.19603v1 Announce Type: new Abstract: Linear probes are widely used in interpretability research and often compared by cosine similarity. The Mahalanobis cosine similarity (MCS) between two directions, which reweights the inner product by test data covariance, is a natural task-aware refinement. Ying et al. (2026) report that a probe's MCS to a reference probe trained on the out-of-distribution (OOD) data near-perfectly linearly predicts the probe's OOD AUROC (R^2 = 0.98). Here, we extend this empirical finding across models, layers, and concept domains, and prove this general phenomenon in closed form: For balanced classes whose projections are Gaussian, OOD AUROC and MCS to the reference probe are linear because both are sigmoid-shaped functions of the probe's signal-to-noise ratio (SNR) on the test data. The theory also predicts when this linearity fails, which we verify empirically. MCS offers a theoretically grounded and empirically effective alternative to Euclidean cosine similarity for comparing linear probes.

阅读与讨论 → 访问原文 →
14.
arXiv (quant-ph) 2026-06-19

Attosecond Path Qubits in High-Harmonic Generation: Classical Dephasing and Trace-Out Decoherence

作者:
A. Marchisio↗C. Granados↗M. F. Ciappina↗O. Cohen↗

arXiv:2606.20372v1 Announce Type: cross Abstract: High-harmonic generation (HHG) is governed by interference between electron trajectories. We propose that the dominant short and long trajectories define an experimentally addressable two-level subsystem: an attosecond path qubit (APQ). We formulate a trajectory-resolved density matrix to identify two distinct coherence-loss mechanisms: classical dephasing from ensemble averaging and quantum decoherence arising from the trace-out of unobserved degrees of freedom. By investigating shot-to-shot fluctuations and unresolved transverse momentum, we demonstrate that while dephasing suppresses coherence through averaging, the ``trace-out'' channel produces mixed states even for fixed driving parameters. We explore how these mechanisms modify APQ purity and show that mode selection and conditioning provide operational routes to isolate them. These results establish a reduced-state framework for diagnosing coherence loss in HHG and for engineering trajectory-based quantum states in attosecond interferometry.

阅读与讨论 → 访问原文 →
15.
arXiv (math.PR) 2026-06-11

Integrated expectile-based measures of inequality

作者:
Ignacio Cascos↗Marco Tarsia↗

arXiv:2606.12333v1 Announce Type: cross Abstract: Expectiles provide a class of asymmetric location functionals that incorporate the magnitude of deviations and admit a natural geometric interpretation. Building on their structural consistency with the convex stochastic order, this paper introduces a family of integrated expectile functionals for measuring risk, dispersion, and inequality. The proposed functionals admit analytical representations as integrals of expectiles across asymmetry levels. For a distinguished subclass of these constructions, a geometric representation is available: the resulting quantities can be expressed as weighted areas of star-shaped sets encoding the distributional asymmetry of a random variable. This approach yields a new class of expectile-based inequality indices, constituting a natural counterpart to classical Gini-type measures while preserving desirable monotonicity and consistency properties. Empirical counterparts are derived in closed form and admit explicit decompositions over finite samples. The framework extends naturally to multivariate settings through directional expectile constructions, leading to measures capable of capturing genuinely joint forms of multivariate dispersion and inequality.

阅读与讨论 → 访问原文 →
16.
arXiv (CS.CV) 2026-06-16

Polyp-D2ATL: Deep Domain-Adaptive Transfer Learning for Colorectal Polyp Classification under Label Distribution Shift

作者:
Sajad Jabarzadeh Ghandilu↗Maryam Sadat Hosseini Azad↗Shahriar Baradaran Shokouhi↗Emad Fatemizadeh↗

Early and highly accurate prediction of colorectal polyps, as an important sign of one of the most dangerous types of cancer, will result in saving more lives. Despite the advancements in colorectal polyp classification, many challenges remain in obtaining an automated polyp prediction system that is able to diagnose the difficult-to-predict polyps accompanied by different features in real scenarios, where the model can handle imbalanced data, label distribution shift, and cross-modality generalization successfully. In this study, we propose Polyp-D2ATL, a novel framework accompanied by a specific training strategy, which mitigates these limitations and effectively predicts the different classes of polyps belonging to the NICE classification. Our extensive experiments on the PICCOLO validation and test sets demonstrate that the proposed Polyp-D2ATL significantly outperforms existing state-of-the-art models across various reliable metrics, achieving an accuracy of 82.38%, a Macro-F1 of 77.49%, and a specificity of 87.47% on the validation set, alongside consistent improvements on the held-out test set which demonstrates the generalization capacity and clinical applicability of the proposed approach.

阅读与讨论 → 访问原文 →
17.
arXiv (CS.CV) 2026-06-17

AlignDrive: Aligned Lateral-Longitudinal Planning for End-to-End Autonomous Driving

作者:
Yanhao Wu↗Haoyang Zhang↗Fei He↗Rui Wu↗Yanhu Shan↗Congpei Qiu↗Liang Gao↗Wei Ke↗Tong Zhang↗

Practical autonomous driving requires models that generalize by reasoning through spatial-temporal possibilities to exclude unsafe outcomes. While state-of-the-art (SOTA) methods use parallel planning architectures, they fail to explicitly couple speed decisions with agent behavior along the driving path, leading to suboptimal coordination. To address this, we propose a cascaded framework that transforms longitudinal planning from an independent prediction task into a path-conditioned reasoning process. On the model side, we introduce an anchor-based regression design that conditions longitudinal prediction on the lateral drive path, and reformulate longitudinal planning as 1D displacement prediction along the path. This reduces geometric uncertainty and sharpens the model's focus on interaction-driven dynamics. On the data side, we introduce a planning-oriented data augmentation strategy that simulates rare safety-critical events by programmatically inserting agents and relabeling longitudinal targets to enforce collision avoidance. Evaluated on the challenging Bench2Drive benchmark, our method achieves SOTA performance with a driving score of 89.07 and a success rate of 73.18%, demonstrating significantly improved coordination and safety. Further evaluation on Fail2Drive confirms strong generalization to rare edge cases where parallel formulations typically fail. Project page:https://yanhaowu.github.io/AlignDrive/.

阅读与讨论 → 访问原文 →
18.
arXiv (CS.CV) 2026-06-16

DenseControl: Instance-Level Controllable Synthesis of Dense Crowd Image

作者:
Juncheng Wang↗Lei Shang↗Wang Lu↗Baigui Sun↗Shujun Wang↗

In this paper, we introduce DenseControl, a novel pipeline for generating dense crowd images. Specifically, DenseControl meticulously positions and sizes each generated instance to align precisely with the predefined coordinates and scales. Based on this, we further allow for control over the background, style, and attributes of instances. The motivation behind DenseControl stems from the observation of two main challenges in synthesizing crowd images: controlling signal embedding and maintaining topological integrity when imparting instance scale guidance. To address these, we first introduce the Isolated Object Embedding (IOE) map, a novel representation that facilitates spatial location control while mitigating the difficulties associated with learning projections for model. Secondly, we propose an Implicit Scale Embedding (ISE) strategy that seamlessly integrates with the IOE map to encode precise scale information. To further enhance the efficacy of combining ISE with the IOE map, we incorporate a Position Shortcut mechanism that enhances cross-attention to alleviate projection challenges. We evaluate DenseControl through two lenses: synthesis quality and applicability in latent applications. Experiments across different control conditions demonstrate DenseControl achieves state-of-the-art results in dense crowd image synthesis. Furthermore, we showcase applications in augmenting crowd analysis under data scarcity, transfer learning, and weather generalization scenes, to highlight the practical utility of DenseControl. The codebase will be released.

阅读与讨论 → 访问原文 →
19.
arXiv (CS.CL) 2026-06-11

Overcoming State Inertia in Full-Duplex Spoken Language Models via Activation Steering

作者:
Cheng-Kuang Chang↗Kai-Wei Chang↗Alexander H. Liu↗James Glass↗

Full-duplex spoken language models (FD-SLMs) enable seamless speech interaction by allowing models to listen and speak simultaneously, yet the internal mechanism by which they coordinate listening and speaking remains underexplored. We analyze the predictive behavior encoded in FD-SLM hidden representations and find that they exhibit stream-specific predictive patterns: during listening, they preferentially predict the incoming user stream, whereas during speaking, they preferentially predict the model output stream. Building on this observation, we show that FD-SLMs dynamically modulate their internal predictive focus between two states: a generative state aligned with model output generation and a perceptive state aligned with incoming user input. However, this modulation can lag behind abrupt changes in conversational context. During user interruptions, the model remains transiently biased toward the generative state before transitioning into the perceptive state, causing it to miss the beginning of the incoming input. We term this delayed internal transition state inertia. To quantify its downstream impact, we introduce the Zero-Buffer Benchmark (ZBB), a diagnostic benchmark for evaluating immediate interruption comprehension when user speech begins abruptly. We evaluate this setting using response correctness and initial-word occurrence rate (IWOR). Finally, we mitigate state inertia through activation steering with a perception vector, a training-free intervention with little additional computational overhead. Across multiple state-of-the-art FD-SLMs, activation steering substantially improves interruption handling; for example, on PersonaPlex, it improves correctness from 28% to 45% and IWOR from 40% to 72% without any fine-tuning.

阅读与讨论 → 访问原文 →
20.
arXiv (CS.CL) 2026-06-12

Does AI Reviewer See the Full Picture? Attacking and Defending Multimodal Peer Review

作者:
Xinyu Zhao↗Rana Muhammad Shahroz Khan↗Zhen Xu↗Zhen Tan↗Tianlong Chen↗

The integration of Large Language Models (LLMs) and Multimodal LLMs (MLLMs) into scientific peer-review workflows introduces novel and significant risks for adversarial manipulation, especially given the multimodal nature of scientific papers where figures, not just text, convey core evidence. This creates a significant gap: current robustness studies on AI peer-review are overwhelmingly text-only. Moreover, the problem is distinct from standard jailbreaking, as a peer-review attack seeks to induce a domain-specific, targeted failure (e.g., "inflate this score") rather than a general safety policy violation, for which no practical defenses exist. To address this, we introduce PaperGuard, the first comprehensive benchmark designed to systematically evaluate and defend AI-generated peer-review against these domain-specific, cross-modal attacks. Our framework is built on three pillars: (1) a new multimodal peer-review dataset spanning multiple scientific domains; (2) a unified suite of attacks, including black-box prompt injections and white-box perturbations, specifically designed to target both text (GCG) and figures (PGD); and (3) a practical defense, motivated by the long-context challenge of academic papers, that uses chunk-based embedding search to efficiently localize and mitigate harmful instructions. Our extensive experiments, conducted across state-of-the-art models, confirm that AI reviewers are pervasively vulnerable. PaperGuard establishes the foundational benchmark, protocols, and actionable defense necessary to pioneer trustworthy, attack-resilient AI-assisted scholarly reviewing.

阅读与讨论 → 访问原文 →
21.
arXiv (CS.CL) 2026-06-18

Trade-offs in Medical LLM Adaptation: An Empirical Study in French QA

作者:
Ikram Belmadani↗Oumaima El Khettari↗Carlos Ramisch↗Frederic Bechet↗Richard Dufour↗Benoit Favre↗

The development of large language models (LLMs) has led to an increased focus on their adaptation to specialized domains and languages, yet the effectiveness of domain adaptation strategies remains unclear. We present a study of medical domain adaptation using French medical question-answering (QA) as a case study. We compare continual pretraining (CPT), supervised fine-tuning (SFT), and their combination across three model families, multiple sizes, and three initialization types, explicitly disentangling adaptation effects from base model choice. We evaluate both multiple-choice (MCQA) and open-ended QA (OEQA) under greedy and constrained decoding using automatic metrics and LLM-as-a-Judge evaluation. For MCQA, CPT+SFT most often achieves the best scores, but gains over SFT are small and frequently not statistically significant, making SFT a strong and cost-effective default. For OEQA, CPT consistently improves overlap-based metrics, while SFT often degrades generation quality; instruction tuning and CPT+SFT are preferred by LLM-based evaluation. Cross-lingual experiments further show effective transfer from French adaptation to English benchmarks. Overall, we provide practical guidelines for selecting adaptation strategies under computational constraints.

阅读与讨论 → 访问原文 →
22.
arXiv (math.PR) 2026-06-11

Instability of a nonlinear oscillator with small friction and small additive noise

作者:
Peter H Baxendale↗

arXiv:2606.11389v1 Announce Type: new Abstract: Let $\lambda = \lambda(\beta,\sigma,a,b)$ denote the top Lyapunov exponent for the linearization along trajectories of the noisy damped non-linear oscillator $\ddot{x}+\beta \dot{x} + ax+bx^3 = \sigma \dot{W}_t$, where $a$, $b$ and $\beta$ are all positive and $\sigma \neq 0$. In 2004 Arnold, Imkeller and Sri Namachchivaya stated without proof that $\lambda(\varepsilon^2 \beta,\varepsilon \sigma,a,b) \sim \overline{\lambda} \varepsilon^{2/3}$ as $\varepsilon \to 0$ with $\overline{\lambda} > 0$. This paper contains a proof of this assertion.

阅读与讨论 → 访问原文 →
23.
arXiv (CS.CL) 2026-06-18

Breaking the Solver Bottleneck: Training Task Generators at the Learnable Frontier

作者:
Lorenz Wolf↗Connor Watts↗Roger Creus Castanyer↗Geoffrey Bradway↗Maxwill Lin↗Augustine N. Mavor-Parker↗Matthew Daborn-Sargent↗

The limiting resource for training agents via reinforcement learning (RL) is increasingly frontier task supply: valid, solvable tasks just difficult enough to train the current model. As reasoning and agentic models improve, fixed task distributions saturate, while naive synthetic generation yields tasks that are trivial, impossible, or ill-posed. Training a task generator with RL to optimize validity and learnability can address this bottleneck, but direct optimization requires repeated solver rollouts per candidate. For software-engineering (SWE) tasks, a single rollout can take tens of minutes; solver-in-the-loop generator training is intractable. We introduce PROPEL, a solver-amortized framework for training task generators at the targeted solve rate. PROPEL trains a lightweight activation probe on a one-time labeled corpus of generated tasks and solver outcomes. The probe predicts target-solver pass rate from a frozen generator reference model and serves as a proxy for solve rate during generator optimization, reducing generator evaluation to a single forward pass. Across math, code, and software-engineering at multiple model scales, PROPEL shifts generation toward the targeted solve rate: for coding, tasks generated at the learnable frontier increase from $10.1\% \rightarrow 20.0\%$ for a Qwen2.5-3B-Instruct solver and from $5.3\% \rightarrow 12.6\%$ for a Qwen2.5-7B-Instruct solver. For SWE, PROPEL increases the share of generations at the targeted solve rate from $9.8\% \rightarrow 19.6\%$ for Qwen3.5-27B on repositories not seen during training of probe and generator.

阅读与讨论 → 访问原文 →
24.
arXiv (CS.AI) 2026-06-19

Beyond Entropy: Learning from Token-Level Distributional Deviations for LLM Reasoning

作者:
Xuanzhi Feng↗Zhengyang Li↗Zeyu Liu↗Haoxi Li↗Yuming Jiang↗Bing Guo↗Jingcai Guo↗Jie Zhang↗Song Guo↗

arXiv:2606.19771v1 Announce Type: new Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced Large Language Model (LLM) reasoning; however, it faces a fundamental optimization instability: uniform token updates precipitate entropy collapse, leading to premature convergence to suboptimal strategies, whereas excessive Shannon Entropy maximization can cause entropy explosion, driving blind exploration toward incoherent reasoning chains. To resolve this dichotomy, we introduce the Independent Combinatorial Tokens (ICT) framework, which shifts the optimization focus from scalar uncertainty to the distributional properties of token logits. By leveraging the Jensen-Shannon (JS) divergence between token logits distributions, ICT identifies tokens with distinctive distributional patterns as critical branching points for guiding effective exploration in LLM reasoning. Our theoretical analysis, grounded in both Shannon and second-order Rényi entropy, proves that selectively updating on these tokens regulates policy concentration: it reduces the overall distribution uncertainty measured by Shannon entropy, while controlling probability concentration captured by second-order Rényi entropy. This dual effect prevents over-concentrated token generation from weakening exploration and effectively stabilizes the training landscape. Empirical results demonstrate that updating only the top 10% of unique tokens on Qwen2.5 (0.5B/1.5B/7B) models yields an average pass@4 improvement of 4.58%, with a maximum gain of 14.9%, over GRPO, 20-Entropy, and STAPO baselines across seven benchmarks spanning math, commonsense, and Olympiad-level problems.

阅读与讨论 → 访问原文 →
25.
arXiv (CS.CV) 2026-06-12

SemanticXR: Low Power and Real-time Queryable Semantic Mapping with an Object-Level Device-Cloud Architecture

作者:
Rahul Singh↗Devdeep Ray↗Connor Smith↗Sarita Adve↗

Semantic mapping is a core service that enables grounded interactions in emerging Extended Reality (XR) applications such as AI assistants and spatial object search. Deploying this capability on mobile XR devices requires a system that is open-vocabulary, real-time, and low-power. Existing approaches are compute-intensive and assume server-class resources. Cloud offloading offers a practical path, but no existing system splits semantic mapping across the device-cloud boundary or manages its communication, execution, and memory footprint. We present SemanticXR, the first device-cloud system for real-time, open-vocabulary semantic mapping and querying under XR power, bandwidth, and memory constraints. Our key insight is to elevate semantically identifiable objects to first-class units of communication, execution, and memory across the device and server. On the server, object-level parallelism and geometry downsampling improve mapping latency, while object-level depth-mapping co-design reduces upstream bandwidth. On the device, an object-level sparse local map with incremental updates and update prioritization enables network-robust querying with bounded memory and downstream bandwidth. Object-level configurable resource usage vs. quality trade-offs let applications and the system adapt mapping to application requirements and operating conditions, respectively. Against a device-cloud baseline with the same perception models, object-level organization improves server-side mapping latency by 2.2X at equal semantic quality. Depth-mapping co-design maintains upstream bandwidth under 2.5 Mbps. On the device, SemanticXR sustains sub-100 ms query latency for up to 10,000 objects even under network drops, supports tens of thousands of objects within 500 MB, and scales downstream bandwidth with map changes, not total scene size. The system adds only 2% device power during normal operation.

阅读与讨论 → 访问原文 →