Knowledge
-
IECG: Weakly-Supervised Interpretability-Enhancing Foundation Model for Electrocardiogram
2026-10-20 12:03:09Huynh Dang Nguyen, Trong-Thang Pham, Anh-Thuan Do, Phuong-Nhi Dinh, Brijesh Patel, Ngan Le, Van NguyenAbstract
Electrocardiography (ECG) is a cornerstone of cardiac diagnosis, yet most machine learning methods for ECG analysis lack interpretability, limiting their clinical adoption. We present IECG, a weakly-supervised ECG foundation model that jointly performs diagnostic classification and abnormality localization without requiring manual abnormal region annotations, overcoming a long-standing bottleneck in interpretable ECG learning. Central to IECG is a weakly-supervised attention strategy that explicitly incorporates clinical priors by deriving abnormality heatmap pseudo-labels from established ECG clinical knowledge. These pseudo-labels guide the model's CLS-token attention through a KL-divergence loss, encouraging it to focus on diagnostically relevant waveform regions. During pretraining, IECG is jointly optimized with clinically guided attention supervision, ECG-text contrastive alignment, and report captioning objectives, enabling it to learn rich multimodal representations while producing localized waveform evidence. By explicitly aggregating features from clinically guided attention regions, IECG grounds its diagnostic predictions in interpretable waveform evidence without sacrificing classification performance. Extensive experiments demonstrate that IECG achieves state-of-the-art abnormal region localization while maintaining strong classification performance, advancing trustworthy foundation models for ECG analysis.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28083
PDF
-
Building Telecom Language Models with RSOD: Domain Data, Specialist Post-Training, Memory-Conscious Distillation, and Networking Evaluation
2026-10-20 10:30:58Anil Srirangapatna Nagesh, Nitin Gupta, Anuj Taneja, Vivek Singh, Shane Eleniak, Theo Zhao, Moulidharan N, Shashank KeshettyAbstract
Telecom language models must combine broad networking knowledge with specialist behavior across routing, broadband services, access technologies, and operational diagnostics while remaining practical to deploy as a single model. We present an end-to-end framework that builds source-grounded synthetic training data from structured networking taxonomies and expert graphs, materializes 212,559 routed training records across fourteen networking routes, trains six frozen specialist LoRA teachers over one shared continued-pretrained backbone, and consolidates their behavior into one deployable student adapter using Routed Specialist On-Policy Distillation (RSOD). In RSOD, the committed student generates each trajectory, the route-assigned specialist teacher-forces the exact native token sequence, compact teacher hidden states are reconstructed through the shared frozen output head, and only the student LoRA is updated. A controlled reconstruction validation shows mean absolute logit error of 7.26 x 10^-6 with 100% top-1 agreement between native and reconstructed teacher logits. On four RTX 6000 Ada GPUs, the completed 2,000-step run produced 16,000 trajectories with zero observed student-version or token- ID mismatches. Under the primary route-conditioned evaluation over 2,850 held-out questions, the raw Meta-Llama-3.1-8B-Instruct model scores 12.8661/25 micro (13.0172/25 macro), the combined routed six-teacher system scores 21.7641/25 micro, and the consolidated RSOD student, released as Llama-CalixCLM v1, reaches 21.7112/25 micro (21.8006/25 macro) on the same five dimension LLM-judge rubric. Under a separate neutral deployment-like prompt, the same student reaches 20.0028/25 micro (20.0778/25 macro). The consolidated student remains within 0.0529 micro-score points of the combined routed-teacher system while retaining a numerically validated and auditable train-serve execution path.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27993
PDF
-
Multimodal-Conditioned Latent Diffusion Models for Spatiotemporal Brain MRI Forecasting
2026-10-20 10:25:35Antonio Scardace, Silvia Ingala, Sebastiano Battiato, Daniele Rav\`iAbstract
Spatiotemporal forecasting of 3D anatomical changes is a critical generative vision challenge for modeling Alzheimer's Disease (AD) progression. While brain MRI forecasting from single-modality data has advanced, other key complementary multimodal biomarkers remain underutilized. In this work, we propose the Multimodal-Conditioned Latent Diffusion Model (MC-LDM), a novel framework for patient-specific brain MRI synthesis. MC-LDM leverages multimodal biomarkers---MRI, tau-PET, Clinical Dementia Rating (CDR), and APOE4---during training, but relies only on baseline MRI, APOE4, and patient metadata at inference, reflecting realistic clinical availability. Our contributions are threefold: (1) a Latent Diffusion Model (LDM) fine-tuned via Low-Rank Adaptation (LoRA) for data-efficient learning; (2) a multimodal conditioning that integrates molecular and clinical signals to model heterogeneous disease trajectories; and (3) auxiliary predictors estimating unobserved multimodal biomarkers from MRI features and patient covariates to condition the diffusion model, removing the need for these assessments at inference. Trained on 1,355 visits from 852 subjects in the Alzheimer's Disease Neuroimaging Initiative (ADNI), MC-LDM reduces average regional volumetric MAE by 18% over the second-best baseline on an independent test set of 664 visits. To our knowledge, MC-LDM represents the first diffusion-based framework to leverage multimodal priors for MRI forecasting. This conceptual advancement enables patient-specific trajectory generation that captures aspects of AD heterogeneity not observable from structural anatomy alone, bridging the gap between rich multimodal supervision and the strict constraints of routine clinical deployment. Code is available at: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.25481
PDF
-
When Do Humans and Machines Truly Complement Each Other? Formal Conditions and Empirical Evidence for Human-AI Synergy
2026-10-20 10:09:59Andrea Campagner, Federico Cabitza, Vincenzo CrupiAbstract
Can humans and artificial intelligence cooperate such that their combination yields greater accuracy than either agent alone? This result, termed \emph{synergy} or \emph{complementarity}, is notoriously difficult to achieve, even when both agents possess significant independent accuracy. In this paper, we present a formal framework to model human-AI interaction through the lens of reliance patterns, which characterize the joint correctness structure of human and algorithmic decisions. We derive mathematically necessary and sufficient conditions for synergy, demonstrating that its emergence depends on the ability of either agent to identify and correct the errors of the other, what we call Human and Algorithmic Correction. Beyond its theoretical contribution, the proposed framework provides a foundation for the design and evaluation of knowledge-based decision support systems, where interaction policies must dynamically integrate heterogeneous sources of expertise. We demonstrate this through the analysis of seven clinical case studies. Our empirical results reveal a ``dominance trap,'' where a strong algorithmic advantage can paradoxically inhibit synergy by eroding human corrective intervention. We also show that synergy can persist even in the presence of cognitive biases, such as automation bias, provided there is functional de-correlation between human and AI errors. These results provide a formal foundation for key concepts in the human-AI systems literature and offer actionable insights for designing interaction-centric AI systems that foster complementarity.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27965
PDF
-
From Proper Scoring Rules and Edge Biases to Mumford-Shah Models: Robust Unsupervised Hyperspectral Image Segmentation
2026-10-20 10:04:28Fabian Conrad, Benjamin BerkelsAbstract
Hyperspectral images provide a rich representation of the underlying spectrum for each pixel, allowing for a pixel-wise classification/segmentation into different classes. As the availability of labeled hyperspectral images is limited and the acquisition of labeled training data is time-consuming, unsupervised segmentation methods are important in hyperspectral image segmentation. Spectral variability, mixed pixels, noise as well as high correlation between neighboring spectral bands make this task challenging. We present a feature-agnostic, unsupervised hyperspectral segmentation framework based on the piecewise-constant Mumford-Shah functional. Our framework turns a strictly proper scoring rule and a family of parametric probability distributions into an indicator function for the Mumford-Shah functional, and we naturally generalize some previous Mumford-Shah based models for image segmentation this way. Our focus will be on heavy-tailed elliptical distributions, the negative log-likelihood score and the score based on the beta-divergence to deal with the heavy outliers in hyperspectral imaging. Using the interplay between the calculus of variations and statistics, we also propose a robustness step to estimate the statistical parameters of these heavy-tailed elliptical distributions. In addition, an edge bias, which contains prior boundary information, is incorporated into both the data fidelity term and the perimeter term of the Mumford-Shah functional. Numerical experiments show that, to the best of our knowledge, our method achieves the highest overall accuracy among fully unsupervised methods on three of the four standard benchmarks (Indian Pines, Kennedy Space Center, Pavia University and Salinas) evaluated on the full scenes. It also improves upon our previous Mumford-Shah based model.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27958
PDF
-
APRIL: Ambiguity-Aware Expert Routing for Respiratory Sound Domain-Incremental Learning
2026-10-20 09:49:56Xinhong Li, Weiping TuAbstract
Automated respiratory sound classification enables continuous, non-invasive respiratory assessment, yet long-term deployment remains challenging due to domain shifts from evolving recording devices and acquisition periods. When privacy and governance regulations prohibit replaying historical data, sequential learning easily overwrites previously acquired knowledge. Preserving dedicated domain experts prevents parameter interference, but selecting the correct expert at inference is difficult under overlapping acoustic distributions and absent domain labels. To resolve this dilemma, we propose APRIL, an exemplar-free domain-incremental framework for RSC. Built on a shared frozen BEATs backbone, APRIL sequentially inherits and isolates low-rank experts to enable new-domain adaptation without altering earlier models. A multi-level prototype router retrieves the top-two candidate experts and dynamically selects the final model by integrating domain-routing posteriors with classification confidence, adaptively weighted by matching ambiguity. We establish standardized device- and temporal-incremental benchmarks on the ICBHI and SPRSound datasets across forward and reversed domain orders. Evaluated over five random seeds, APRIL consistently achieves the highest final-session average Score among compared continual learning approaches, reaching 64.4% and 65.4% on ICBHI, and 83.1% and 82.2% on SPRSound under forward and reversed sequences, respectively. Crucially, average forgetting remains under 1.0% across all protocols, demonstrating robust domain adaptation and long-term knowledge retention without data replay. Overall, APRIL offers a reliable and privacy-preserving paradigm for deploying continuous respiratory sound classification across real-world healthcare settings.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27942
PDF
-
ACE-Compression: An Ability-Centric Evaluation of Compressed Large Language Models
2026-10-20 09:26:20Jiahui Ning, Liang Chen, Yuchen Liang, Yi Zhang, Fenglei FanAbstract
Model compression makes large language models deployable, but today's evaluation tells an incomplete story. Standard practice measures retained task performance and stops there. This misses a critical fact: compression perturbs the distributed knowledge embedded across billions of weights, and task scores alone cannot reveal which underlying abilities survive and which quietly collapse. To solve this question, we introduce ACE-Compression, an ability-centric evaluation framework. Instead of asking only "does the compressed model still answer correctly?", we ask "which of its core abilities remain intact?" Our framework probes four deployment-facing dimensions: procedural reasoning and constraint execution, knowledge grounding and truthfulness, social cognition and affective understanding, and persona fidelity and stylistic consistency. Across more than 35 compressed checkpoints from six open-weight LLMs-covering quantization and pruning-we find that compression redistributes ability profiles. Coarse, local, and surface-level signals often persist, while fine-grained, long-tail, and context-dependent abilities become fragile. Quantization consistently outperforms pruning in retention, but the practical meaning of that gap depends entirely on which ability an application requires. Based on our findings, deployers can choose compression methods by the ability profile their application needs, not just by a leaderboard average. And algorithm designers gain diagnostic signals to build compression methods that preserve what matters. We believe that ACE-Compression turns compression from a black-box accuracy drop into an interpretable intervention-one we can measure, understand, and ultimately improve.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27913
PDF
-
PURE: Refining Feed-forward 3D Reconstruction with Unlabeled Post-training
2026-10-20 09:25:44Yaolong Zhu, Peng Li, Jiahao Lu, Chengfeng Zhao, Mengfei Li, Chao Wang, Wenyue Chen, Zekai Gu, Chengwei Ren, Ronggang Wang, Yuan LiuAbstract
Feed-forward 3D reconstruction models have achieved remarkable progress through large-scale training and strong learned priors. However, they remain fragile when applied to dynamic in-the-wild videos, where non-rigid motion, occlusions, and illumination changes are prevalent. A key limitation lies in their reliance on expensive depth and camera pose annotations, which hinders scaling to diverse real-world dynamic data and restricts generalization. To address this issue, we propose \textbf{PURE}, an unlabeled post-training refinement framework that improves depth and camera estimation without requiring ground-truth supervision. Instead of relying on annotations, PURE elegantly distills robust structural knowledge from state-of-the-art 2D priors into 4D feed-forward architectures. PURE leverages relative depth constraints, affine-invariant geometry distillation, and pose refinement based on feature correspondences to stabilize geometric predictions under challenging dynamic conditions. Extensive experiments on both synthetic and real-world dynamic benchmarks demonstrate that PURE delivers consistent improvements across multiple state-of-the-art architectures. Furthermore, our data-efficient paradigm effectively mitigates domain shifts, successfully unlocking the architectural limits of these base models in challenging, in-the-wild dynamic scenes.
Abstract (translated)
URL
https://arxiv.org/abs/2610.14698
PDF
-
Staged Continual Adaptation of Multimodal Foundation Models for Japanese Financial Documents
2026-10-20 09:19:36Genshin Kakimoto, Atsushi YanagisawaAbstract
Staged post-training of multimodal foundation models chains vision-language alignment, reasoning distillation, and domain tuning, yet how each phase trades capabilities against the others remains under-characterised. Tracking an 8.4B-parameter model from an un-fine-tuned baseline through three training phases on Japanese financial documents (evaluated on five text-input benchmarks), we find the phases interact non-additively: math-only distillation overwrites the financial knowledge acquired earlier even as procedural math skill survives, and the domain phase repairs more than it installs. A skip-phase ablation is best for domain knowledge but forfeits numerical-reasoning gains, so per-phase contributions are not decomposable and the best checkpoint is task-dependent, with implications for checkpoint retention and scheduling rather than added stages.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27904
PDF
-
EndoWorld-JEPA: Motion-Guided Global-Local Latent World Modeling for Endoscopic Video Representation Learning
2026-10-20 09:11:49Jie Wang, Shilong Yang, Tao Zhou, Juan Yu, Rong Qu, Jianfeng Ren, Zhicheng Zhang, Linlin ShenAbstract
Endoscopic video contains state transitions induced by camera displacement, deformable tissue, and instrument interaction, yet existing pre-training methods treat dynamics implicitly. Inspired by action-conditioned joint-embedding predictive world models and by the observation that endoscopic optical flow contains both shared image-plane displacement and localized residual motion, we reformulate passive endoscopic video pre-training as causal motion-guided latent-state prediction. We present EndoWorld-JEPA, to our knowledge the first latent world-model pre-training framework designed for general-purpose endoscopic video representation learning. In place of recorded actions, calibrated poses, or robot kinematics, the model conditions prediction of the next latent state on motion observed up to the current state. Its Optical Motion-guided module separates dominant displacement from residual motion and converts the resulting evidence into an ordered temporal motion condition and spatial local-motion weights. Complementary global and motion-local predictors use these outputs: the temporal condition modulates both latent transitions, whereas the local weights define an auxiliary motion-local state that extends prediction from scene-level evolution to spatially selective dynamics. Patch-wise motion weighting emphasizes localized change while retaining anatomical context, and past-motion-window dropout randomly withholds the complete motion condition to reduce dependence on unreliable motion. This global--local formulation learns predictable visual--motion dynamics in representation space without reconstructing stochastic appearance artifacts. Evaluation shows that a common pre-trained checkpoint achieves strong transfer across four downstream benchmarks, supporting motion-guided global--local latent prediction as an effective pre-training paradigm for endoscopic video.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27890
PDF
-
Hallucinations of Large Language Models in Cross-Language Code Translation: Phenomena and Mitigation
2026-10-20 08:50:39Yali Du, Yunji Zhang, Ming LiAbstract
Code translation converts source code from one programming language to another while preserving its functionality. Although Large Language Models (LLMs) have improved code translation, their outputs still contain hallucinations: generated code that conflicts with the source program, violates target-language facts, or introduces contextually inconsistent dependencies. These failures are especially harmful in translation because small semantic deviations can silently change program behavior. Existing hallucination studies mainly focus on natural-language generation or code generation, leaving the forms, distribution, and mitigation of hallucinations in cross-language code translation insufficiently characterized. This paper presents a systematic empirical study of LLM hallucinations in code translation. Across three benchmark datasets, six language pairs, and seven LLMs, we construct a taxonomy with three major categories: Semantic Consistency Conflicts, Factual Knowledge Conflicts, and Project Context Deviations. We manually annotate 13,919 hallucination cases and analyze how hallucination types vary across models, datasets, and language directions. Guided by the taxonomy, we further propose HaluCodeTran, a mitigation framework that combines BM25-based retrieval of relevant translation examples with taxonomy-aware reflective refinement. The study provides an execution-tested characterization of code-translation hallucinations and a practical baseline for building more reliable LLM-based code migration tools.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27868
PDF
-
Agentic discovery of interpretable car-following laws from multi-source trajectory data
2026-10-20 08:25:21Jiamei Xu, Jie Sun, Keshuang Tang, Jian SunAbstract
Car-following (CF) models is a fundamental component of microscopic traffic simulation and underpin the analysis of collective traffic dynamics, while its modeling becomes more challenging due to diverse human driving behaviors and the increasing presence of automated vehicle. Analytical CF models offer interpretable representations of longitudinal vehicle interactions, but their prescribed functional structures may limit the representation of heterogeneous driving behavior. This study proposes an agent-centered, physics-aware framework for the empirical discovery of CF laws from multi-source trajectory data. The framework iteratively combines agent-guided hypothesis generation, informed by traffic flow knowledge and state-dependent reconstruction errors, with evolutionary symbolic search and physics-aware multi-objective evaluation. Using NGSIM US-101, highD, and exiD, the framework discovers the Adaptive Equilibrium Response Model (AERM), which couples state-dependent relative-speed response with spacing-dependent equilibrium-speed restoration. AERM admits analytical equilibrium speed-spacing and flow-density relationships; its stability conditions are derived for the zero-delay formulation, and platoon simulations examine disturbance propagation. Evaluated against seven classical CF models on Shanghai naturalistic driving data, NGSIM I-80, and Waymo trajectories, AERM attains the lowest spacing and speed reconstruction errors under trajectory-specific calibration, reducing spacing RMSE by 9.11%, 7.59%, and 21.73% relative to the strongest baselines, and remains competitive under dataset-level shared parameters. These results demonstrate the fitting flexibility and analytical tractability of the discovered CF law, providing an interpretable basis for microscopic traffic simulation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27836
PDF
-
RAG-NIMBF: A Non-Invasive, Modular Benchmarking Framework for Automated RAG Configuration Analysis
2026-10-20 08:08:08Niclas Hart, Nicolas WeegerAbstract
Retrieval-Augmented Generation (RAG) has become the dominant paradigm for grounding the outputs of large language models (LLM) in external knowledge. However, practitioners are faced with a vast configuration space comprising chunking, embedding, retrieval, reranking, prompting and generation models, and there is limited guidance on how these components interact. We present RAG-NIMBF, a modular benchmarking framework that attaches to existing RAG pipelines as a non-invasive evaluation layer. It evaluates the configuration space through controlled ablation with end-to-end experiment tracking. The evaluation combines LLM-based answer quality scores, retrieval correctness measures, and runtime indicators. This study validates the framework across multiple generation models on the SQuAD dataset. This process reveals non-obvious interactions and demonstrates how the experiments are tracked. The findings indicate that the retrieval balance of relevance and diversity substantially enhances the relevance and coverage of retrieved context in comparison to a conventional similarity search, with MMR raising nDCG@5 by +4 to +11 percentage points across the swept models. More detailed prompt instructions meaningfully increase how faithfully generated answers stick to the retrieved evidence, especially for capable and reasoning models (+10 to +13pp on Qwen3.6-27B, GPT-4o-mini and Qwen3.5-think, but negligible on the 20B generator). Overall, RAG-NIMBF transforms the configuration space of RAG into reproducible, comparable experiments. By attaching to existing production pipelines in a non-invasive way, it enables practitioners to systematically evaluate and improve deployed systems without disrupting existing workflows. This replaces ad hoc choices with evidence-based decisions.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27816
PDF
-
Toward Agentic AI for the 6G Physical Layer: From Algorithm Discovery to Verified Baseband Silicon
2026-10-20 07:56:10Eric Mittag, Lukas Brechtel, Maximilian Stark, Hans D. SchottenAbstract
6G will be designed faster than it can be built: AI-driven search already proposes physical-layer algorithms at machine speed, while carrying a single one onto embedded baseband silicon remains specialist work. That silicon imposes hard memory and numeric limits, but also exposes cost structures a GPU lacks, and a design flow that never runs on the target sees neither. We present an agentic workflow in which an LLM coding agent implements algorithms on a commercial baseband SoC, optimizes them down to agent-written vector assembly, and validates them on silicon. The agent is guided by a knowledge base of measured platform facts, checked at every step against evaluators it is instructed never to edit, and metered per phase. We showcase the workflow on an AI-discovered orthogonal time frequency space (OTFS) detector portfolio on the open 6G4All platform. It yields twelve kernels that run cycle-exactly on silicon, their optimized versions at 18k-405k cycles (30-660 $\mu$s) per frame. It also yields a new detector shaped by the chip's measured costs that needs 13.9% fewer cycles than the port of the discovered detector while matching its error rate, with a measured edge our evaluation cannot statistically resolve, a structure that a hardware-blind search objective had no reason to find. What is not specific to OTFS is the workflow: an LLM agent exploring an implementation space under real hardware limits, with the cost of every choice measured rather than assumed.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27797
PDF
-
Are Linguistic and Arithmetic Circuits Dissociated in LLMs?
2026-10-20 07:52:47Riku Kisako, Tatsuki Kuribayashi, Ryohei SasanoAbstract
Large language models (LLMs) exhibit both knowledge of language, rendered by, for example, grammatical judgment ability, and abilities to perform tasks through language, such as math-word problem solving. Motivated by neurolinguistic studies of their neural dissociation in human brains, in this study, we investigate the overlap of their circuits in large language models (LLMs). Specifically, we examine the overlap and causal effects between the circuits extracted for linguistic and arithmetic tasks, using controlled materials and circuit discovery methods from the mechanistic interpretability field. Our experiments with five LLMs show that circuit similarity is mostly explained by task domains (linguistic vs. arithmetic), rather than by their input and output vocabularies or by the syntactic similarities between natural-language sentences and arithmetic expressions. This finding suggests that LLMs may process linguistic and arithmetic information using largely separate neural circuits, highlighting an ability-based modular organization of circuits within these models.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27790
PDF
-
Beyond Calibration: Temporal LLM Confidence Assessment Across Languages and Local Settings
Abstract
Large language models are expected to estimate their own uncertainty, with confidence signals used to guide retrieval, adaptive compute allocation, routing, and selective abstention. We investigate whether confidence signals measured at different temporal stages of problem solving encode distinct forms of epistemic information and remain reliable beyond English-centric settings. Inspired by classical meta-memory theory, we distinguish anticipatory pre-solve confidence (Feeling-of-Knowing, FOK) from retrospective post-solve confidence (Judgement-of-Learning, JOL). We evaluate these signals across two frontier LLMs, multiple reasoning domains, translated Marathi benchmarks, and a newly constructed benchmark of genuinely local Marathi factual questions. Across settings, the cheaper pre-solve signal remains weakly calibrated, while post-solve confidence generally provides stronger epistemic alignment despite its additional inference cost. Confidence behaviour transfers surprisingly well under direct language translation, with calibration trends largely preserved between English and Marathi, but degrades substantially on local Marathi knowledge, suggesting that linguistic transfer alone is insufficient for region-specific information needs. Finally, hallucinated responses maintain persistently elevated confidence after answer generation, highlighting important limitations of current LLM self-assessment mechanisms.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27702
PDF
-
VLA-Retain: Benchmarking Cross-Dimension Effects of Vision Tuning in VLAs
2026-10-20 06:17:00Borui Xu, Han Chen, Bingsheng He, Ming Wu, Jason Zeng, Michael HeinrichAbstract
Vision-language-action (VLA) models transfer knowledge from large-scale pretraining into robot control, but their manipulation performance remains sensitive to changes in observation conditions, which recent work addresses by fine-tuning the visual module on target-domain data. However, it remains unclear how such adaptation affects robustness beyond the target condition, as existing robustness benchmarks only evaluate models without fine-tuning and adaptation studies only report the target gain. Measuring this cross-dimension effect requires separating target gains from off-target changes, isolating adaptation effects from architectural differences, ruling out generic forgetting, and comparing tuning choices. Therefore, we introduce VLA-Retain, a benchmark that compares five open VLA models, tuned on Camera-view or Sensor-noise data, with their own base checkpoints on LIBERO-Plus dataset spanning seven perturbation dimensions. Our study reveals four findings. First, targeted vision tuning is not local, and its target gain does not predict retention. Mean off-target gain is negative in 8 of 10 model--domain configurations. Under Camera-view tuning, X-VLA gains $69.4$ percentage points on the target but loses $4.6$ points on average elsewhere, whereas OpenVLA-OFT gains $39.3$ and $1.4$ points, respectively. Second, these regressions are not explained only by performance loss on the same tasks without perturbations. Third, broader training coverage raises overall gain over target-specific training but does not eliminate dimension-level regressions. Fourth, tuning strategies are model-dependent, as DELTA-style feature preservation raises X-VLA's mean off-target gain to $+1.2$ points but lowers OpenVLA-OFT's to $-23.1$. Overall, these findings show that target-domain success alone is insufficient and highlight cross-dimensional retention as a key criterion for targeted VLA adaptation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27698
PDF
-
A Unified Attack Framework against Statistical Generative Watermarks for Large Language Models
2026-10-20 05:12:12Romina Omidi, Nima Naderloui, Yuan Hong, Binghui WangAbstract
Watermarking has emerged as a key tool for identifying text generated by large language models (LLMs), with current schemes embedding statistical signals into the generation process to enable later detection. In this work, we identify a shared weakness across a broad class of popular LLM watermarking methods, including KGW (and its variants), GumbelSoft, DipMark, ITS, and SynthID. We show that these schemes exhibit a fundamental link between their detection scores and the token probability shifts they introduce during generation. This link, which we call score-ratio alignment, allows an attacker with only black-box query access to the model--and no knowledge of secret keys or model weights--to produce text that evades detection while remaining fluent and natural. We prove that the resulting attack is optimal across all covered watermark families, and show that the weakness is intrinsic: any watermark designed to maximize statistical detection power necessarily induces the very structure that attackers can exploit. We evaluate our attack on seven popular generative LLM watermarking schemes, three LLMs, and tasks, which reduce detection rates to at most 0.05 across all settings while preserving text quality. Our findings show that current watermarking methods can be reliably defeated under realistic attack conditions, highlighting the need for fundamentally different watermarking designs that decouple detection scores from token probability shifts.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27653
PDF
-
Corpus Quality Improves, Retrieval Utility Does Not: An Ablation of Preprocessing and Refinement in a Speech-to-QA Pipeline
2026-10-20 05:11:32Wonsup Shin, Jingu KimAbstract
Practitioners building training corpora with a pipeline of language models keep a preprocessing or refinement (post-process) stage when switching it on raises the corpus quality. We ablate a production pipeline that turns conference speech into question-answer pairs. The preprocessing stage rewrites the transcript, and the refinement stage filters the generated pairs. Switching each on and off gives four corpora from the same 40 Korean medical and finance sessions. On the corpus itself, quality rises as the stages switch on. Four LLM judges and six human domain experts rank the four corpora the same way, and the fully refined corpus leads by +0.18 and +0.22 on a 1-5 scale. We then ask whether that gain carries downstream. We supply each corpus to six models at inference and measure retrieval accuracy. The ranking does not hold. Enabling the transcript rewrite lowers retrieval accuracy by 0.061, the only stage effect that survives correction for multiple comparisons. That stage also reads the whole session, so it almost doubles the cost of one. With both stages on, accuracy does not change. An equivalence test we declared in advance confirms this. The corpus itself is clearly valuable. It lifts retrieval accuracy from 0.231 to 0.748 in every model and domain we test. More than a third of its pairs carry knowledge that neither of two frontier models produces on its own. On our data, the null does not come from weak corpora. A rising score on a rubric of this kind did not predict what the corpus holds for a model.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27652
PDF
-
Speak vs. Know: Balancing Behavior and Knowledge Distributions of Large Language Models
2026-10-20 05:07:35Yang Liu, Masahiro Kaneko, Chenhui ChuAbstract
Recent work reports that the distribution verbalized directly in the output text aligns far better with human opinion distributions than the next-token log-probability distribution of LLMs. However, the verbalized distribution reflects what LLMs know about a group's opinions (knowledge distribution), while what they actually output at deployment is governed by their log-probability distribution (behavior distribution). We find that when prompting LLMs to predict the opinion of a specific group, the most likely answer under the verbalized distribution differs from that under the log-probability distribution for 37% of World Values Survey questions, averaged over three LLMs. Thus, LLMs' knowledge does not match their behavior. This matters for evaluation: a verbalized distribution can score well against a group while its most likely answer is one the model never outputs, so the score no longer describes the model. The alignment scores for these two distributions reflect only the model's behavior or its knowledge, separately. We therefore propose a learnable combination of the two distributions that follows the most likely answer of the behavior distribution while preserving the shape of the knowledge distribution, with a single hyperparameter trading off behavioral fidelity against distribution quality. Our experiments on three opinion survey datasets and six models show that the balanced distribution improves behavioral consistency over the verbalized distribution from 71.6% to 95.6% on average, while the total variation distance to the human opinion distributions rises only slightly from 28.8% to 33.8%, and it remains effective under temperature scaling and cross-dataset settings.
Abstract (translated)
URL
https://arxiv.org/abs/2610.27648
PDF
