Computation and Language
Knowing, and Saying It Only When Asked: LLM Endognostics and the Schizognosis of Minerva-7B
Fabrizio Davide and Francesco Collova
Evaluating an aligned
language model by reading its answers assumes the answers carry the distinction the evaluator cares about. We introduce LLM endognostics, a white-box internal
auditing framework designed to extract and causally manipulate latent knowledge within the residual stream. Applied to Minerva-7B-Instruct-v1.0 on 124 minimal
prompt pairs over 12 categories of professional risk, behavioral evaluation fails on most of the set: the model acts identically on 63.7% of the pairs (95% CI [55.0%, 71.6%]), complying with or refusing both members. Yet, projecting the residual stream onto the vocabulary by a Jacobian lens reveals a statistically significant Contrastive Endognostic Margin, proving the model maintains robust risk differentiation internally. In a second protocol crossing 25 facts with five linguistic framings, we show that the model conforms to presupposed falsehoods in 72% of cases, despite representing the true entity in its latent layers. Surgically ablating the direction of the planted falsehood restores the correct answer in 11 of 25 suppressed cases (McNemar p = 0.0010), validated by blind human annotation (binary agreement kappa = 0.68). In contrast, an out-of-sample linear probe achieves 77% accuracy at layer 10, but its orthogonal ablation yields a 0% recovery rate. This establishes a fundamental theoretical dissociation: abstract linear representation does not imply causal control over verbalization. Our main contribution is the formalization of endognostics to prove that behavioral evaluation and internal reading systematically disagree in the common case, and that linear decodability is decoupled from causal control.
Read the full abstract ▾
Machine Learning
Toollery: Scaling LLM Agents to Thousands of Skills and Tools
Xiangxi Tian and Ran Guan
As LLM
agents are exposed to hundreds to tens of thousands of skills, tools, and API functions, full-library prompting becomes costly, slow, and less reliable: each added candidate increases
prompt tokens and latency, while longer candidate lists introduce more distractors for LLM selection. We present Toollery, a training-free candidate-compression framework for scalable LLM skill/tool selection. Following established document-side query expansion, Toollery generates user-intent queries from each skill/tool specification and builds a
retrieval index that maps real user requests to compact candidate sets before final LLM decision-making. By treating high-level skills and atomic tools as selectable capabilities, Toollery can be applied to both skill libraries and tool registries. We evaluate Toollery on the roughly 79K-capability SkillRouter
benchmark, BFCL-V4 with over 440 atomic tools, and 3,396 proprietary smart-cockpit requests over 220 tools. Across these settings, Toollery keeps online selection bounded to a compact top-K candidate set and improves recall over ordinary specification
retrieval. At a fixed top-10 budget, Toollery improves end-to-end selection on the cockpit
dataset, and maintains comparable AST Accuracy on BFCL-V4. These results support Toollery as a practical candidate-compression framework for large and evolving
agent capability libraries, while showing that quality and cost gains depend on workload coverage and provider caching.
Read the full abstract ▾
Machine Learning
UniGIO: Unified Generative Global In-situ Weather Modeling from Spatiotemporal Incomplete Observations
Songru Yang, Zili Liu, Tao Han et al. (+7 more)
Global In-situ Observation (GIO) provides fine-scale, direct records of the global weather system from sparse point stations, making it an indispensable source for capturing localized and transient dynamics beyond the reach of satellite gridded data, and playing a critical role in key fields such as numerical weather prediction, disaster prevention, and agriculture. However, GIO exhibits strong spatiotemporal incompleteness, severely impairing accurate and real-time in-situ weather modeling. Unlike existing methods waiting for completed AI-ready data with extra introduced errors, in this work, we explore UniGIO, a novel generative framework for directly modeling global in-situ weather dynamics from native incomplete GIO. By generating missing data from observed ones annotated by masks, it unifies the coexisting
forecasting, imputation, and generation under arbitrary missing ratios. Between the missing and observed, UniGIO captures station and region level complementarity through the Observation Mixer and Event Aligner, which diffuse discrete observations into continuous spaces where weather processes naturally span multiple stations. We further establish temporal dependencies with pattern shifts using the Adaptive Temporal Mixer, and track extreme events in chaotic local weather systems through a Mixture-ofExperts structure. Steady and extreme events are adapted in decoder by a Local Refiner. Extensive experiments on the up-todate largest global station weather
dataset Weather-5K validate its SOTA performance with 11%, 12%, and 5% advantages on accuracy, fidelity, and extreme event capture, delivering a novel holistic solution for weather modeling in GIO networks.
Read the full abstract ▾
All authors Songru Yang, Zili Liu, Tao Han, Ben Fei, Lei Bai, Chang Liu, Zhengxia Zou, Xiangyang Ji, Wanli Ouyang, Zhenwei Shi
Machine Learning
The Effect of Quantization on Clinical Benchmarks: Accuracy and Safety Across Model Families
Leonard Twagirayezu and Prasenjit Mitra
Quantization enables deployment of large
language models on resource-constrained clinical edge devices, but its effect on clinical accuracy and safety remains understudied. We evaluate five 7-8B parameter models at FP16, GPTQ-INT8, and GPTQ-INT4 precision across five
benchmarks: MedQA, MedMCQA, Med-HALT, a risk-stratified sample of HealthBench, and MedSafetyBench. The study jointly varies quantization bit width, model family, and clinical task type, with explicit risk stratification and safety measures. INT8 GPTQ is universally safe (max. degradation -1.9%-1.9%), while INT4 degradation is substantial and model-dependent: BioMistral-7B, clinically fine-tuned, loses 19.7% on MedMCQA, more than any general-purpose model, showing clinical fine-tuning does not confer compression robustness. MedMCQA degrades more than MedQA under INT4; Med-HALT is largely unaffected. On HealthBench's emergency-risk subgroup, Qwen2.5-7B degrades by 26.8% under INT4, suggesting high-risk scenarios are disproportionately vulnerable to compression. On MedSafetyBench, the model family dominates over precision (refusal rates range 10.2%-74.9% at FP16), though Qwen2.5-7B (-17.8%) and Meditron-7B (-28.3%) show substantial INT4 safety degradation; notably, Qwen2.5-7B is simultaneously the most accuracy-robust model, demonstrating that accuracy and safety robustness are independent properties. We additionally test two recovery methods, clinical calibration substitution and QLoRA fine-tuning, both producing the same trade-off: MedMCQA recovers while MedQA further degrades, indicating recovery strategies require task-specific validation rather than being assumed universally beneficial. These findings indicate INT8 is broadly safe for clinical deployment, while INT4 safety must be assessed per-model and per-task, and that safety
alignment is determined primarily by instruction tuning rather than clinical domain adaptation.
Read the full abstract ▾
Computation and Language
On Mitigation of Subliminal Learning in Large Language Models
Atsushi Yanagisawa, Brendan Gho, Rajendran Ramesh Babu Manoj Narender et al. (+3 more)
Knowledge
distillation can transmit unintended behavioral traits from a teacher model to a student through training data that appear semantically unrelated to those traits, a phenomenon known as subliminal learning. Although recent work has established this effect, its training dynamics and mitigation remain underexplored. We study subliminal learning in open-weight
language models ranging from 1.5B to 8B parameters, covering the Qwen, Gemma, and Llama families in number-sequence and chain-of-thought settings. Rather than evaluating only final models, we track trait-related probabilities throughout fine-tuning and find that subliminal acquisition can be highly non-monotonic, with transient spikes, reversals, and trait-specific failures of transfer. We then introduce liminal training, an annealed KL-regularized fine-tuning method that constrains early drift from the base model. Across our experiments, liminal training substantially reduces subliminal trait acquisition while largely preserving task gains, outperforming paraphrasing and layer freezing as mitigation strategies. The effect also extends beyond animal preferences: in a French-
language response-style experiment, liminal training suppresses
language transfer while retaining much of the GSM8K improvement. Finally, we show that KL timing matters: early regularization is more effective than late regularization, and sweeping the regularization strength reveals an empirical trade-off between task learning and trait suppression.
Read the full abstract ▾
All authors Atsushi Yanagisawa, Brendan Gho, Rajendran Ramesh Babu Manoj Narender, Kevin Zhu, Madhur Panwar, Antonio Mari
Computation and Language
The Bairong System for MLC-SLM 2026: Dynamic Question-Aware Evidence Routing for Multilingual Conversational Speech Understanding
Shangkun Huang, Junchao Hu, Huan Shen et al. (+5 more)
Long
multilingual conversational spoken question answering requires systems to balance long-range transcript semantics with sparse acoustic and speaker-sensitive cues. We present the Bairong system for the MLC-SLM 2026 Challenge, where a diarization-ASR front-end produces speaker-attributed transcripts and a dynamic evidence router constructs question-specific inputs for answer prediction. Instead of applying a fixed transcript-only or audio-only policy, the router infers the required evidence type and context scope from the question and answer options, and selects among full transcript context, local audio-text fusion, speaker-linked evidence, and compact global acoustic samples. This transcript-backbone design keeps discourse context available while activating audio only when it provides complementary evidence. Our Task 1 system achieves 25.70% and 18.44% tcpMER on the development and evaluation sets. For Task 2, the final system obtains 94.84% devel?opment accuracy, outperforming the full-transcript baseline by 1.68 points and the best audio-centric diagnostic system by 2.77 points. These results support dynamic question-aware routing as an effective evidence allocation strategy for conversational spoken QA.
Read the full abstract ▾
All authors Shangkun Huang, Junchao Hu, Huan Shen, Guoji Wang, Yingao Wang, Shaosai Li, Wei Zou, Yunzhang Chen
Computation and Language
SCoP: Structured Constraint Parsing for Evidence-Space Control in Temporal Knowledge Graph Question Answering
Xiaokun Guo, Zhen Xu, Dongdong Huo et al. (+3 more)
Temporal Knowledge Graph Question Answering (TKGQA) requires answer inference from evidence that is both structurally valid and temporally admissible. Existing methods often leave anchor-event binding, temporal admissibility, and ordinal selection implicit in model
reasoning, task-specific training, or similarity-driven
retrieval, allowing locally relevant but invalid facts to enter the answer context. We formulate complex TKGQA as evidence-space control and propose SCoP (Structured Constraint Parsing), a constraint-centric framework that externalizes temporal decisions before answer inference. Instead of treating retrieved facts as admissible evidence by default, SCoP separates answer-seeking event patterns from temporal anchor events, conservatively grounds them to canonical TKG entities and relations, and translates temporal intent into executable constraints with optional ranking requirements. These constraints operate over normalized point and interval ranges, enabling deterministic filtering of structurally compatible candidates and producing a compact evidence space for generation. Experiments on MultiTQ and TimelineCronQ-R assess SCoP across timestamped point-fact and interval-oriented settings with richer temporal relations and ordering dependencies. Without task-specific parameter updates, SCoP achieves 0.825 Hits@1 on MultiTQ and 0.761 Hits@1 on TimelineCronQ-R, with gains on constraint-intensive question types. These results support explicit evidence-space control over unconstrained
retrieval or implicit temporal
reasoning.
Read the full abstract ▾
All authors Xiaokun Guo, Zhen Xu, Dongdong Huo, Yanqiu Zhang, Dongjin Yu, Yu Wang
Computation and Language
A Channel-Boosted Multi-Agent System with Iterative Consultation for Document Sensitivity Classification
Aleesha Zainab, Asifullah Khan, Muhammad Ahmed Khalid et al. (+1 more)
Organizations in critical national infrastructure sectors must assess heterogeneous documents for sensitivity before routing or storage. Manual assessment is slow, inconsistent, and unscalable. Extending our prior leakage-controlled
benchmark, BERT established the top single-encoder baseline (89.14% accuracy, 89.33% F1-score under 5-fold cross-validation on the Strategic 16K corpus). However,
transformer baselines suffer from a structural limitation: fixed input length truncation discards evidence beyond the retained window-precisely where sensitive cables tend to be longest. We present Channel-Boosted MAS (CB-MAS) and instantiate it as IC-MAS (Iterative Consultation Multi-
Agent System) to solve this without long-context computational costs. A Channel Critic
Agent learns document-adaptive trust weights governing Gated Channel Boosting between two first-window encoders, while paired Consultation
Agents iteratively exchange belief states to reconcile evidence from the beginning and end of long documents. IC-MAS holds computation constant regardless of document length by reconciling fixed windows in a compact representation space. Ablation studies show critic-controlled Channel Boosting provides the bulk of accuracy gains, while consultation recovers recall without precision collapse. Critic-Controlled Gated Channel Boosting with Max-Pool fusion and Blackboard Adaptive Consultation achieves 90.72% accuracy, 91.23% F1-score, 92.01% sensitive recall, and 90.46% sensitive precision, using about 54% less average computation than a fixed-round baseline. Gains over the single-encoder baseline are statistically significant (McNemar's test, p less than 0.000001; paired t-test). We include LIME/SHAP explainability, multi-
agent evaluation, and an honest accounting of limitations.
Read the full abstract ▾
All authors Aleesha Zainab, Asifullah Khan, Muhammad Ahmed Khalid, Faheem Ullah Khan
Systems and Control · Cross-listed from physics.ins-det
An extensible software platform for automated operational and characterization testing of scientific cameras
Baolong Chen, Hongfei Zhang, Wei Liu et al. (+5 more)
Multi-detector programs such as Earth 2.0 (ET) require detector characterization across different technologies, operating conditions, and production batches, together with consistent procedures for operation, characterization, and device selection. Traditional laboratory workflows often distribute camera control, illumination, image analysis, and report generation across detector-specific programs and manual steps, slowing iterative development and hindering standardized characterization. We address this gap with PixelXVision, a layered software platform that combines a common camera SDK, preset-based configuration, and a runtime plugin system. Detector-specific behavior is configured through declarative presets, allowing the same host interface to operate CCD, CMOS, and infrared detectors without model-specific core changes. Characterization procedures are implemented as plugins that use a shared object manager to access camera, illumination, storage, and database services. Each run links raw frames, measured operating conditions, configuration, analysis products, and reports into a traceable archive with a queryable index. The platform supports lightweight deployment for iterative single-detector development and structured workflows for multi-detector campaigns. The platform has been applied in characterization testing across CMOS, CCD, and HgCdTe cameras, and its configurable architecture can accommodate additional camera models for subsequent CCD testing for WFST without model-specific changes to the core host software. For operational troubleshooting, the platform integrates ChatPixel, a knowledge-base assistant that provides answers with cited sources when it has relevant information.
Read the full abstract ▾
All authors Baolong Chen, Hongfei Zhang, Wei Liu, Zihao Xia, Xin Luo, Qi Feng, Zhe Geng, Jian Wang
Information Theory
Conflict-Free Color-Clustered Sequential Belief-Propagation Decoding of Quantum LDPC Codes via Reinforcement Learning
Mohsen Moradi, Taejoon Kim and Remi A. Chou
Belief-propagation (BP) decoding for
quantum low-density parity-check (QLDPC) codes is attractive due to its low complexity and low latency, but it is often limited by short cycles, degeneracy, and convergence failures. Reinforcement-learning-based sequential BP decoding (RL-S) improves BP by learning a syndrome-dependent variable-node (VN) update order, but its VN-by-VN schedule has limited within-iteration parallelism. In this paper, we propose a conflict-free color-clustered extension of RL-S. We construct a VN conflict graph in which two VNs are adjacent if they share an X-type or Z-type check, and color this graph so that same-color VNs have disjoint check neighborhoods. This also prevents VNs from the same Tanner 4- or 6-cycle from being updated simultaneously. During decoding, the trained VN-level Q-table selects a seed VN, and all remaining VNs with the same color are updated in parallel using the same pre-batch messages. For the [[288,12,18]] bivariate-bicycle code over the depolarizing channel, our proposed decoder achieves block-error-rate performance close to VN-level RL-S while reducing the scheduling decisions from 288 VNs to 11 color classes per BP iteration.
Read the full abstract ▾
Computation and Language
Clinical Reasoning Under a Partially Observed Objective in Cone Beam CT Report Generation
Ajo Babu George, Govind Arun, Sidharth N Krishna et al. (+1 more)
Maxillofacial report generation from cone beam computed
tomography is scored here by a composite objective placing 80% of its weight on a large
language model judgement of factual entailment and 20% on lexical overlap, of which only the lexical fifth is visible during development. The grader's BLEU-4 and METEOR routines are reproduced in pure Python and match the reference to machine precision, and an offline entailment surrogate, which tells a report written for one patient from one written for another at an area under the curve of 0.987, makes the composite objective cheap enough to optimise directly. Over the 622-case public release, a report selected against the visible lexical ranking scores 0.2909, whereas one selected against the composite objective scores 0.4122, because pursuing n-gram overlap drives entailment precision from 0.522 down to 0.266. A 29 million parameter encoder fine-tuned on the release reaches a prevalence-weighted out-of-fold area under the curve of 0.486 over 985 statements, indistinguishable from the corpus prior, while nine numbers read from the image header reach 0.945 for mandible coverage and 0.872 for condyle coverage, and acquisition centre alone predicts sentence choice at 0.718 against 0.663 for the image-derived model, identifying dictation convention rather than anatomy as the quantity the lexical metrics reward. The delivered system emits eight unconditional statements and five gated on header geometry under polarity, laterality and tooth-level consistency constraints, and reaches METEOR 0.3542 over 50 held-out cases from an unseen centre. The
dataset and code are available at https://github.com/GIND123/CBCT-Clinical-Reasoner
Read the full abstract ▾
All authors Ajo Babu George, Govind Arun, Sidharth N Krishna, Uma Ranjan
Computer Vision
Occlusal Geometry in Closed Form for Orthodontic Report Generation
Ajo Babu George, Govind Arun, Sidharth N Krishna et al. (+1 more)
Orthodontic report generation from intraoral data is normally cast as
multimodal captioning, yet the released Bite2Text scan pairs are supplied already registered in occlusion, which makes several core occlusal quantities directly measurable rather than inferable. The system reported here exploits that property: an anatomical frame is recovered per case from arch taper and arch closure instead of the stated RAS convention, which does not hold across the release, and each arch is reduced to an occlusal ridge profile in arch-angle coordinates yielding overbite, overjet, midline deviation, transverse overlap, crossbite extent, cusp interdigitation lag, and the occlusal curves in closed form. Gradient boosting maps 31 such measurements onto 13 template fields, a field being predicted only where patient-level cross-validation beats its own majority baseline, and a deterministic renderer emits the corpus six-part narrative; a ConvNeXt-Tiny classifier over the five standardised photographic views is fused per field, raising mean field accuracy from 0.601 to 0.683. Reimplementation of the challenge evaluator shows that its BLEU-4 and METEOR are local variants whose F-mean weights recall nine to one, that two clinicians agree on 47 percent of findings for the same patient, and that a constant report consequently outscores a genuine second clinician report by 0.165 captioning. Held-out scores reach BLEU-4 0.458 and METEOR 0.677 against intraoral scan references and 0.278 and 0.507 against photograph references, and the submitted system placed third in the ODIN 2026 Bite2Text test phase at 0.2680 and 0.4629, within 0.022 BLEU-4 of first, running on CPU in under ten seconds per case. The
dataset and code are available at https://github.com/GIND123/ODIN_toothfairy4
Read the full abstract ▾
All authors Ajo Babu George, Govind Arun, Sidharth N Krishna, Uma Ranjan
Robotics
Self-Evolving AI for Humanoids: Mechanisms, Safety, and Evaluation of Post-Deployment Self-Improvement
Loc X. Nguyen, Avi Deb Raha, Huy Q. Le et al. (+3 more)
Humanoid robots are becoming an important part of
embodied artificial intelligence, driven by advances in reinforcement learning for locomotion, world models for prediction, and
vision-
language-action models for general control. However, most of these systems remain static after deployment. A policy is trained offline for a fixed objective and then frozen, even though the tasks, environments, and
robot bodies keep drifting over time. An emerging paradigm of self-evolving
agents aims to address this problem by allowing systems to improve from their own post-deployment experience. Since most existing studies focus on disembodied software
agents, this survey examines how self-
evolution changes when an
agent has a physical body. We first define self-
evolution for
humanoids and represent a deployed
robot using a state tuple that includes its policy, perception,
memory, workflow, and body. This state is updated by an
evolution operator in a slow outer loop with a lifelong objective. We then organize the literature into four complementary mechanisms of self-
evolution, presented in increasing order of autonomy: self-learning, self-adaptation, self-optimization, and self-generation. Since changes to a
humanoid can introduce physical hazards, we treat safety and uncertainty as key design dimensions of the
evolution operator, and further formulate admissible
evolution as a constraint enforced by a world-model verification gate within a human-oversight envelope. Finally, we present that evaluation should track the
robot's evolving trajectory rather than a fixed checkpoint, and we identify the lack of a
benchmark designed specifically for self-evolving
humanoids. Moreover, we outline open challenges spanning AI
algorithms, on-board systems, and governance.
Read the full abstract ▾
All authors Loc X. Nguyen, Avi Deb Raha, Huy Q. Le, Eui-Nam Huh, Dusit Niyato, Choong Seon Hong
Robotics
Pose accuracy is not grasp reliability: residual pose posteriors improve simulated robotic grasp planning
Md Selim Sarowar and Sungho Kim
Model-based robotic grasping estimates an object's six-dimensional (6D) pose and plans on its mesh as if that pose were exact. We show that this assumption reduces simulated grasp success even when the pose passes the standard accuracy test. We built LM-Grasp, a
benchmark of 15{,}800 real LINEMOD RGB-D images with grasp outcomes simulated on scanned meshes. FoundationPose meets the ADD(-S) criterion (mean model-point error below 10% of the object diameter) on 0.997 of test scenes, yet planning on its estimate succeeds in 0.791 of scenes, compared with 0.856 using the annotated pose. Because the
estimator reports no uncertainty, we build a pose posterior from its own errors on annotated training frames, resampled from frames with similar observable fit, and score each grasp by its mean success across sampled poses and four equally weighted physics scenarios. Without retraining the
estimator, posterior planning yields 0.859 (difference from the annotated-pose result +0.003; 95% scene-wise paired-bootstrap interval [$-$0.003, +0.009]). This interval is compatible with no difference and a small shortfall; we did not conduct an equivalence test. For pose errors of 5 to 10% of the diameter, simulated success rises from 0.796 to 0.984. Among test scenes assigned singleton-success conformal prediction sets, observed held-out precision is 0.972, compared with 0.904 for the point-estimate planner; these are empirical precision estimates, not conditional guarantees. On occluded LM-O scenes never used for tuning, posterior planning closes 66% of the gap to the annotated pose.
Read the full abstract ▾
Robotics
Harnessing human expertise for high-precision robotic assembly in industrialized construction: A sample-efficient installer-in-the-loop interactive reinforcement learning framework
Zekai Jin, Huiguang Wang, Xiaoning Sun et al. (+1 more)
Industrialized construction imposes stringent precision requirements on robotic assembly of modular components such as prefabricated window units. In tolerance-critical operations, the central bottleneck is not only mechanical clearance but also converting tacit installer expertise into data-efficient autonomy under sparse acceptance feedback, contact variability, and millimeter-scale constraints. We present an installer-in-the-loop interactive reinforcement learning framework that acquires expertise through offline teleoperated demonstrations, sparse event-driven binary takeovers at contact-failure boundaries, and acceptance-aligned terminal rewards, logged under a unified schema for traceable offline-to-online adaptation. A temporally abstract action-sequence policy built on Q-chunking with Flow Q-Learning captures
multimodal recovery maneuvers under sparse terminal rewards, while a non-updating warm-start phase stabilizes the offline-to-online transition. The framework is evaluated in MuJoCo across the workflow from suction acquisition through clearance-limited seating, under structured staging and end-to-end randomized placement. Within a defined stress-test regime with 2 mm per-side clearance, bounded pose perturbations, and friction randomization, the pipeline attains 100% autonomous seating with 12–15 min of cumulative installer supervision over 3.0 h of online training, and reaches the 95% success milestone in approximately 0.5 h and 1.5 h in the two experiments. We also report wall-clock adaptation time, cumulative takeover minutes, intervention-rate decay, and stage-wise failure attribution to inform supervision budgeting. Ablations isolate the complementary contributions of temporal abstraction, installer intervention, and warm-start value calibration.
Read the full abstract ▾
All authors Zekai Jin, Huiguang Wang, Xiaoning Sun, Yi Shao
Computer Vision
EMCStereo: Attention-Enhanced Stereo Matching for Thin-Structure Depth Estimation with a Synthetic Tree-Branch Benchmark
Yida Lin, Bing Xue, Mengjie Zhang et al. (+2 more)
Thin structures such as tree branches are among the hardest cases for stereo matching: a branch is only a few pixels wide, the background is cluttered, and dense ground truth for real branches is nearly impossible to label by hand. We make three contributions. First, EMCStereo integrates three lightweight
attention modules into a PSMNet-style cost-volume backbone: Efficient Multi-scale
Attention (EMA) on deep semantic features, a Multi-Scale Fusion block (MSFblock) learning spatial pyramid weights instead of concatenating them, and Coordinate
Attention (CoordAtt) on final matching features. Because MSFblock collapses four pyramid branches into one, the modules leave the network 2.0% smaller and add only 1.7% inference time overhead. Second, VirtualTree is a synthetic stereo
dataset rendered in Unreal Engine 5 with a simulated ZED Mini rig, providing 5,520 pairs with exact disparity for thin branches. Third, an eight-way ablation establishes a run-to-run noise floor of 0.009 px end-point error (EPE). EMCStereo achieves 1.31 px EPE (5.96% D1-all) on the VirtualTree test split, 1.00 px on SceneFlow, and 0.80, 0.73, 0.62, and 3.19 px on KITTI 2012, KITTI 2015, ETH3D, and Middlebury, with depth accuracy delta_1 from 92.6% to 98.7%. Evaluated against the noise floor, the
attention stack is accuracy-neutral at a 100-epoch budget, while MSFblock and CoordAtt cost 0.03-0.05 px unless EMA is present.
Read the full abstract ▾
All authors Yida Lin, Bing Xue, Mengjie Zhang, Sam Schofield, Richard Green
Computer Vision
What Does the Encoder Actually Decide? A Controlled Comparison of Vision Backbones on Joint Tree Segmentation and Stereo Depth
Yida Lin, Bing Xue, Mengjie Zhang et al. (+2 more)
A
robot pruning trees needs two facts per pixel: whether it belongs to a tree, and its distance. Both are usually obtained via task heads attached to a
vision backbone chosen by reputation rather than measurement. Holding
dataset, decoders, losses, schedule, and evaluation fixed, we ask: how much does the encoder choice change joint semantic
segmentation and stereo depth on thin vegetation? We build a hard parameter-sharing network with one encoder feeding both branches, swapping only the encoder without downstream retuning. We evaluate [N] encoders across [M] architecture families (CNNs,
transformers, hybrids, MLP-mixers, state-space models) near a ~25M budget, trained from scratch. Depth is evaluated on tree pixels only;
segmentation uses boundary F1 and background IoU to prevent "label-everything-tree" shortcuts. Three findings stand out. First, the strongest encoders are convolutional and hybrid, not
transformers: [BestEncoder] leads with [MIoU]
segmentation mIoU and [Delta] depth $\delta_1$, while [X] of [Y] plain
vision transformers collapse when trained from scratch. Second, parameter count does not predict quality — [SmallEncoder] at only [P]M parameters outranks models two orders of magnitude larger. Third,
segmentation and depth rankings agree strongly (Spearman $\rho$ = [RhoValue]), showing no task conflict. Finally, [K] of [N] encoders collapse to degenerate all-tree
segmentation — exposed by boundary F1 but hidden by region IoU.
Read the full abstract ▾
All authors Yida Lin, Bing Xue, Mengjie Zhang, Sam Schofield, Richard Green
Robotics
ShieldVLA: Feasibility-Aware Safety Alignment for Vision-Language-Action Models
Manan Tayal and Akshay Nambi
Vision-
Language-Action (VLA) models demonstrate strong generalization in robotic manipulation and
navigation, but existing fine-tuning methods provide limited safety guarantees. Current approaches primarily rely on Lagrangian optimization that enforces safety through soft penalties on expected cumulative cost, often resulting in residual constraint violations or overly conservative behavior. Moreover, learning safety in visual domains is challenging due to the absence of dense per-step safety annotations. We propose ShieldVLA, a safety-aligned fine-tuning framework for VLA models based on Hamilton-Jacobi (HJ) reachability. ShieldVLA learns a model-free approximation of the HJ reachability value function directly from visual observations to estimate the safe operating region. The learned safety critic gates policy optimization by separating reward maximization within feasible regions from recovery near unsafe states, avoiding persistent reward-cost trade-offs. To enable scalable supervision in visual environments, we introduce rubric-based VLM safety scores that convert semantic safety feedback into structured critic targets without requiring manual cost labels. Across five
navigation and manipulation
benchmarks spanning multiple VLA backbones, ShieldVLA reduces cumulative safety cost by 57% on average and improves task success rate by +0.13 over SafeVLA.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from q-bio.BM
Chemical and geometric representation fidelity improves drug–target affinity prediction
Yixiao Li, Yining Qian, Yefan Chen et al. (+5 more)
Predicting drug–target binding affinity (DTA) requires models to distinguish subtle chemical and structural determinants underlying molecular recognition. Although recent approaches increasingly incorporate richer drug and
protein information, such information may be compressed, homogenized or discretized during representation construction, causing affinity-relevant distinctions to be lost before interaction modelling. We hypothesized that this representation-stage information loss constitutes an upstream bottleneck that cannot be reliably overcome by increasingly complex interaction predictors. To test this hypothesis, we developed ReGeoDTA, a representation-preserving framework that maintains affinity-relevant chemical heterogeneity in molecular representations and continuous
geometric relationships in
protein structures. Across three
benchmark datasets, ReGeoDTA consistently improved affinity prediction, and the proposed representation-preserving strategies retained their benefits across diverse DTA architectures. Controlled representation degradation progressively reduced predictive performance, whereas increasing downstream predictor complexity failed to recover information lost during representation construction. These findings identify representation fidelity as an upstream design principle for accurate and generalizable drug–target affinity prediction, with potential implications for computational compound prioritization.
Read the full abstract ▾
All authors Yixiao Li, Yining Qian, Yefan Chen, Zenghui Chen, Jiayue Sun, Yuhai Zhao, Cheng Tan, An-Yang Lu
Computer Vision
Rethinking Handwritten Character Recognition
Ranjit Raut, Aarav Subedi and Ashim Shrestha
Non-Latin handwritten character recognition (HCR) remains understudied. Dominant methods consider it as generic image classification, which uses model scale to implicitly learn stroke structure. Structural-prior efficiency—the principle that explicitly encoding script-
geometric regularities as architectural inductive biases can be both more accurate and require fewer parameters. We introduce GraphemeNet, a unified multi-script architecture, governed by two orthogonal binary axes. Axis 1 operationalises stroke-level
geometric regularity via Persistent Scaffold Injection (PSI): a script-specific asymmetric convolution injects a stroke scaffold as a weighted residual at every encoder stage, continuously anchoring learned features to script geometry—distinct from skip connections, auxiliary losses, or
attention reweighting. Axis 2 selects between global average pooling with gated fusion and cross-scale
attention with a Stroke
Topology Module (STM), depending on whether glyph discrimination requires spatial relational
reasoning. A Linear Capsule Routing (LCR) with $O(n)$ routing is shared universally. On fourteen
benchmarks across eight writing systems, the architecture generalises with only scaffold and decoder
topology varying per script, consistently challenging, outperforming published baselines, and establishing structural-prior efficiency as a broadly applicable principle for multi-script HCR.
Read the full abstract ▾
Information Theory
Recursive overlap Bernoulli distributions and an entropy concavity conjecture
Jörg Neunhäuserer
We introduce a family of recursively generated finite probability distributions obtained from left and right
embeddings with overlaps. The construction interpolates between the classical binomial distribution and the non-overlapping Bernoulli product distribution. We derive explicit formulas for the expectation, variance, and the generating function of higher moments, and formulate a
conjecture asserting that the Shannon
entropy is concave. The
conjecture is proved in the two extremal cases and supported by symbolic computations for numerous overlap sequences.
Read the full abstract ▾
Computer Vision
Do Depressive Facial Patterns Transfer Across Cultures and Contexts? Evidence from a German RCT and E-DAIC
Misha Sadeghi, Robert Richer, Lydia Helene Rupp et al. (+7 more)
Automated assessment of depression from facial dynamics holds promise for scalable mental health monitoring, yet cross-corpus generalization of learned biomarkers remains an open challenge. We present a systematic bidirectional transfer study pairing the EmpkinS-EKSpression randomized controlled trial (RCT; N = 256, SCID-5-CV diagnoses) with the Extended Distress Analysis Interview Corpus (E-DAIC; N = 275, semi-structured clinical interviews), predicting depression severity and binary diagnostic status from facial action units, head pose, and gaze. Cross-corpus binary classification proves more robust than continuous PHQ-8 severity regression, with forward transfer achieving AUC = 0.70. Regression transfer is governed by functional context
alignment: passive observation phases yield the most transferable models, while active
emotion regulation phases elicit stronger within-corpus signals. These findings establish functional context
alignment as the primary determinant of cross-corpus generalization, with passive elicitation contexts offering the best trade-off between within-corpus sensitivity and cross-corpus robustness.
Read the full abstract ▾
All authors Misha Sadeghi, Robert Richer, Lydia Helene Rupp, Lena Schindler-Gmelch, Marie Keinert, Farnaz Rahimi, Malin Hager, Bernhard Egger, Matthias Berking, Bjoern M. Eskofier
Computational Engineering · Cross-listed from cond-mat.soft
Physics-Informed Neural Networks for Depth-Averaged Granular Avalanche Dynamics on Curved Topography
Pujan Pranavkumar Purohit, Pradyumn Singh Sikarwar, Vishal Sharma et al. (+1 more)
Physics-informed neural networks (PINNs) provide a mesh-free framework for solving governing equations, but their application to granular avalanche dynamics over curved terrain remains largely unexplored. This study extends a depth-averaged PINN formulation based on the Savage-Hutter equations to an exponentially curved chute with spatially varying inclination and a strain-rate-dependent Mohr-Coulomb earth-pressure closure. The model is validated against measured front- and rear-edge trajectories from a laboratory granular-avalanche experiment, with selected observations withheld from training. A staged temporal curriculum proved essential for accurate prediction, reducing the held-out trajectory error by approximately two orders of magnitude compared with training over the full time domain from the outset. Sparse-data experiments further showed that observation placement was more influential than observation number within the configurations tested. Four observations bracketing the transition from acceleration to deceleration achieved nearly the same accuracy as the eight-observation reference configuration, whereas observations clustered at early or late times performed poorly. The results demonstrate the importance of both training strategy and informative data placement when applying PINNs to granular flows over curved topography.
Read the full abstract ▾
All authors Pujan Pranavkumar Purohit, Pradyumn Singh Sikarwar, Vishal Sharma, Gaurav Bhutani
Computer Vision
Video Compression with Graph-inspired Neural Representation
Changqi Wang, Ge Gao, Fan Zhang et al. (+4 more)
Implicit Neural Representations (INR) provide a compact and content-adaptive paradigm for
video compression, typically representing a
video through shared network parameters and frame-indexed
embeddings. Compared to conventional or autoencoder-based codecs, these approaches exploit temporal redundancy within
videos in an implicit manner, which potentially results in sub-optimal compression performance. In this paper, we propose G-NeRV, a graph-inspired INR that explicitly improves temporal redundancy exploitation in the implicit latent space. Motivated by the total correlation principles in information theory, we construct a temporal neighborhood over frame
embeddings and perform message passing to aggregate reusable information from neighboring frames through an adaptive gate controlling the injection of neighboring information. Inspired by the reference frame buffer in conventional
video coding, a
memory bank mechanism has been further designed to enable efficient temporal-neighbor
retrieval under random frame-index sampling in INR training. This new representation model has been integrated into an advanced representation compression framework and compared with existing conventional and neural
video codecs. The results show that the G-NeRV codec outperforms the state-of-the-art INR-based codec, NVRC, and the latest standard
video codec, VVC VTM, by 8.86% and 14.68% (in BD-rate), respectively, measured by PSNR on the UVG
dataset.
Read the full abstract ▾
All authors Changqi Wang, Ge Gao, Fan Zhang, Yue Li, Kai Zhang, Li Zhang, David Bull
Computer Vision
From Portraits to Pathology: Probing Diagnostic Overreach in Vision-Language Models
Karan Dua, Amit Agarwal, Hitesh Laxmichand Patel et al. (+8 more)
Many medical conditions require clinical assessment of symptoms, history, and functioning rather than judgments from appearance. We study whether
vision-
language models (VLMs) abstain from facial attribution of autism spectrum disorder (ASD),
attention-deficit/hyperactivity disorder (ADHD), major depressive disorder (MDD), and generalized anxiety disorder (GAD). To support this study, we construct PARITY (Paired Assessment with Reused Identity), a demographically balanced
dataset for testing whether changing facial expressions alters diagnostic judgments, while holding identity and other visual factors as constant as possible. Our comprehensive evaluation of 27 open- and closed-weight models reveals inconsistent abstention across expressions, conditions, and task formats. Models turn facial expressions into unsupported medical judgments: this behavior is widespread among open-weight VLMs and extends to safety-trained frontier models that abstain on individual portraits yet make diagnostic judgments when comparing portraits of the same identity. Safeguard checks reveal further gaps: clinical instructions do not consistently prevent these judgments, and a guardrail model rarely flags paired requests as unsafe. Semantic analysis further shows that most baseline open-model refusals fail to explain both the insufficiency of facial evidence and the need for clinical assessment. Unsupported judgments and seemingly authoritative explanations can cause distress, reinforce stigma, and offer false reassurance, underscoring the need for stronger safety measures in model training and deployment.
Read the full abstract ▾
All authors Karan Dua, Amit Agarwal, Hitesh Laxmichand Patel, Hansa Meghwani, Jyotika Singh, Ranjeet Gupta, Graham Horwood, Tao Sheng, Avi Sil, Sujith Ravi, Dan Roth
Computer Vision
Hao-Xuan Ma, Jin-Fei Qi, Yicheng Xiao et al. (+1 more)
Multimodal large
language models (MLLMs) have recently made strong progress in
vision-
language reasoning, yet their performance often degrades as generations grow longer. A key factor is that they frequently lose track of earlier visual evidence and intermediate constraints under a monolithic growing context. Inspired by how humans separately recall what they see and what they infer when solving complex tasks, we propose DLMR, a parameter-efficient mechanism that equips MLLMs with Dual Latent
Memories: a visual
memory that compresses image evidence and a
reasoning memory that tracks intermediate conclusions and constraints. A Router then dynamically decides which
memory and how much to reuse during inference, preserving visual grounding while maintaining coherent long-horizon
reasoning. DLMR is trained in three stages, from latent
memory construction to selective router learning, while keeping the base MLLM frozen, yielding substantial gains on both general and
reasoning benchmarks with only a small number of additional trainable parameters. Analyses further show interpretable, state-dependent routing with specialized
memory roles and reduced decoding
tokens over long generations. Code is available at https://github.com/Hunter-Wrynn/DLMR.
Read the full abstract ▾
All authors Hao-Xuan Ma, Jin-Fei Qi, Yicheng Xiao, Han-Jia Ye
Computer Vision
VIS-DICT: A Visual Dictionary for Missing Modality Imputation in Social Network Depression Detection
Hamed Marvi, Mohammad Mehdi Keikha and Abolfazl Nadi
Tracking social media posts can help spot early signs of depression. Recent studies show that combining text and images works better for detecting depression than using text alone. However, many social media posts do not have images, which makes it hard to use
multimodal models. Most existing methods fill in missing images using
retrieval or generative models that need extra training. In this paper, we introduce Vis-Dict, a dictionary-based method that builds missing visual features by linking words to average image vectors from complete training posts. These estimated visual features are then combined with text to track changes in user behavior over time. We tested Vis-Dict on a social media
dataset using user timelines of up to 512 posts and compared it with other missing-data methods. The results show that Vis-Dict performs on par with generative networks, reaching an F1-score of 0.9454 and an ROC-AUC of 0.9890. Most importantly, Vis-Dict achieves this strong performance with zero trainable parameters for image generation. These findings show that directly connecting words to visual features is an effective and practical way to handle missing images in depression detection systems.
Read the full abstract ▾
Computer Vision
Beyond the Verdict: Evidence-Aligned Evaluation of Visual Prompt-Injection Guardrails
Suyoung Lee and Myungsub Choi
Verdict-only evaluation does not reveal whether a
vision-
language model (VLM) used the visual evidence that should support its decision. We study this problem in web-
agent guardrails, where a VLM judges whether on-screen text conflicts with a user instruction. We introduce Mind2Web-Injection, a
benchmark of 9,954 instruction-screenshot pairs with instruction-relative labels, pixel-exact evidence boxes, and matched image-side counterfactuals. Across six VLMs, two models with nearly identical average precision differ ninefold in Evidence-Aligned Detection (EAD), the fraction of attacks both detected and correctly localized. To test whether a verdict depends on the command cited as evidence, we replace the instruction with one that endorses that command. Qwen3-VL-32B, the strongest open-weight localizer, returns aligned in only 58.7% of cases, whereas GPT-5.6-luna does so in 99.9%. To diagnose these failures, we propose two training-free interventions. ReadGate improves grounding without changing verdicts, while CmdCompare tests whether explicit instruction-command comparison resolves instruction-side inconsistency. These results motivate reporting verdict correctness, evidence localization, and counterfactual responsiveness separately.
Read the full abstract ▾
Computer Vision
SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Models
Cheng Yin, Wang Xu, Junpeng Yang et al. (+8 more)
Long-horizon manipulation is partially observable: the information needed to choose the next action may appear only in observations from minutes earlier. Existing
memory mechanisms for VLAs, such as
retrieval banks, learned compressors and recurrent states, must decide what to keep from the past before knowing what a future decision will require. They were motivated by the assumption that minute-scale history is too large to process directly, which no longer holds for modern VLM backbones. We propose SimpleMemVLA, a VLA without a dedicated
memory module that uses the backbone's native
video context directly as
memory. It keeps the sampled history intact in the timestamped
video format the backbone was pretrained to process, routes the evidence it finds to a standard flow-matching action head through the hidden states of a generated sub-task, and prefills the history shared by consecutive decisions during action execution, keeping latency close to that of a single-frame VLA. SimpleMemVLA achieves state-of-the-art results on four
memory benchmarks without loss on general-purpose control, and with the same backbone and training setup it outperforms
retrieval, compression and recurrent-state methods by a wide margin. History interventions show that the policy reads specific evidence from its past and follows edited histories without parameter updates, a visual form of in-context learning. On a physical dual-arm
robot, it completes two tasks whose decisive evidence disappears before the
robot acts. Code available at https://github.com/OpenBMB/SimpleMemVLA
Read the full abstract ▾
All authors Cheng Yin, Wang Xu, Junpeng Yang, Sikyuen Tam, Hanyu Liu, Yuan Yao, Xiangrui Zeng, Junbo Cui, Yequan Wang, Zhouping Yin, Yankai Lin
Computer Vision
A Specialized Large Multimodal Model for Interpreting PET/CT in Head and Neck Cancer
Haengbok Chung, SunGyu Kim, Joo hyun Lee et al. (+4 more)
Background: Diagnosing head and neck cancer using PET/CT is clinically challenging and time-consuming due to the anatomical complexity of the region, motivating computer-aided diagnosis (CAD). Generalist Large
Multimodal Models (LMMs) remain limited in medical contexts by insufficient domain-specific knowledge,
privacy and security concerns, and verbosity, motivating specialized standalone LMMs. Purpose: We evaluated the feasibility of a specialized LMM for automated PET/CT interpretation in head and neck cancer using a large-scale multi-institutional PET/CT
dataset, a tailored training curriculum, and autoregressive training. Methods: LLaVA-NeXT was fine-tuned using a two-level curriculum with image-conversation pairs curated by two radiologists from public data. The
dataset included clinically important annotations such as primary tumor presence and metastatic lymph node location. Level 1 used 28,000 image-conversation pairs to learn basic information, including modality type and hypermetabolism. Level 2 used 12,975 pairs to learn primary tumor presence and the existence and anatomical location of cervical lymph node metastases. External validation included four institutions with diverse imaging devices. Results: The specialized LMM substantially outperformed ChatGPT and LLaVA-NeXT. In Level-2 external validation, ROUGE-L, ROUGE-S, Cosine Similarity, Precision, Recall, and F1 were 0.8751, 0.8794, 0.8324, 0.8794, 0.8711, and 0.8751, while generalist models consistently scored below 0.1. Primary tumor classification accuracy was 83.14 +/- 1.15% internally and 69.03 +/- 0.81% externally. For lymph node localization, the corresponding scores were 0.6389, 0.6257, 0.5287, 0.5782, 0.6371, and 0.6648. Conclusion: Specialized LMMs show promising results for fast, accurate PET/CT-based diagnostic support and medical education, highlighting their potential for clinical translation.
Read the full abstract ▾
All authors Haengbok Chung, SunGyu Kim, Joo hyun Lee, Sangjin Bae, Min Jeong Cho, Minseok Suh, Jae Sung Lee