Papers

Visual brand language is the set of visual properties that convey brand identity for a product. What is the impact of visual brand language on a person's ability to recognize and understand the functional identity of an object? Using an empirically supported modeling framework based on the JIM model of object recognition and the LISA model of analogical inference, we simulated the impact of visual brand language on object recognition, the allocation of attention, and retrieval of functional information about objects. Our simulations predict that brand information captures attention and can slow recognition of an object's functional category, with greater degrees of branding causing larger effects. These results have potential implications for the usability and experience of designed objects.

Video understanding is moving beyond closed-context perception toward open-world evidence exploration, a paradigm formalized as Video Deep Research (VDR). However, existing multimodal search agents primarily target static images, and the current VDR benchmark relies on text-centric retrieval that discards crucial visual information. To address these limitations, we propose VideoSearcher, a closed-loop agentic framework that empowers Vision-Language Models with multi-tool reasoning for VDR. VideoSearcher unifies temporal localization, spatial focusing, and multimodal search within a single reasoning trajectory, enabling agents to progressively ground visual clues, retrieve relevant evidence, and synthesize answers. To optimize knowledge-intensive reasoning trajectories, we propose Bi-branch Sequence Policy Optimization (BiSPO), a reinforcement learning algorithm that decouples tool-invocation optimization from answer-accuracy optimization. This design provides stable learning signals for both evidence-grounded reasoning and purposeful tool use. Furthermore, we construct VideoSearch-QA, the first benchmark designed to evaluate open-world video information grounding and multimodal search-based reasoning. Extensive experiments demonstrate that VideoSearcher significantly outperforms prior open-source agentic baselines across various search-oriented and multimodal understanding benchmarks.

Quantitative 3D spatial reasoning from egocentric RGB-D video is a critical capability for next-generation wearable assistants. Yet existing benchmarks do not reflect the challenges of handling (1) natural egocentric video, (2) posed RGB-D video inputs, and (3) challenging quantitative 3D spatial reasoning Q&A. To fill this gap, we introduce R3D-Bench (Reasoning in 3D), a benchmark of 3,033 quantitative spatial reasoning questions across 15 types – spanning multiple-choice, distance-based, and volumetric reasoning questions – built on top of 57 egocentric video sequences from Aria Digital Twin. To set a strong baseline on this dataset, we introduce R3D, a model-agnostic spatial tool-calling framework. In contrast to existing approaches that directly embed 3D information into the model's input representation, R3D constructs a 3D scene from video using segmentation and depth-lifted object representations. It provides this information to an LLM through eight composable spatial tools. On R3D-Bench, R3D with Qwen3-VL 235B achieves 73.5% mean relative accuracy, substantially outperforming the best depth-enabled baseline (CuTR+Tools, 61.9%) and the best RGB-only baseline (Gemini 3 Flash, 46.5%).

Magnetic particle imaging (MPI) enables real-time, radiation-free tracking of magnetic nanoparticle-coated instruments, making it highly suitable for interventional procedures. This study proposes a harmonic-aware transformer framework that directly predicts catheter tip positions from raw MPI voltage signals, eliminating the need for image reconstruction and reducing computational latency. The framework incorporates frequency-domain preprocessing to isolate the 2nd to 8th drive-field harmonics, enhancing the signal-to-noise ratio while preserving motion-relevant features. A transformer architecture with six encoder layers and eight attention heads is employed to learn spatio-temporal dependencies across the three receive axes (x, y, z) for accurate three-dimensional position estimation. The model is trained on simulated MPI signals and evaluated on real in vitro datasets under standard, bending, and heartbeat-like motion conditions. The proposed method achieves sub-millimeter localization accuracy, with a minimum L2 error of 0.103 +/- 0.092 mm and mean absolute errors (MAEs) of 0.039 +/- 0.046 mm, 0.054 +/- 0.049 mm, and 0.060 +/- 0.044 mm along the (x, y, z) axes, respectively, for the bending dataset. Across all datasets, the MAE ranges from 0.165 mm to 0.655 mm, demonstrating consistent performance. The optimized inference achieves a latency of 0.55 ms per frame and a throughput of approximately 1800 frames per second, confirming real-time capability. Compared with conventional MPI-guided approaches relying on image reconstruction, the proposed framework provides improved accuracy, reduced latency, and enhanced robustness under complex motion conditions. These results highlight the potential of harmonic-aware transformer models as efficient and scalable solutions for real-time catheter localization in interventional MPI.

The ability to efficiently navigate high-dimensional spaces to identify optimal candidates-whether designing targeted molecular structures or generating images with preferred attributes-stands as a central pillar of modern scientific progress. Recent tree-based samplers enable highly scalable inference-time search by eliminating the reward-gradient bottleneck inherent to particle-based methods. Despite their potential, existing tree-based samplers are fundamentally constrained by the massive number of function evaluations (NFEs) required for node valuation, crippling their utility in exploration-heavy tasks. Furthermore, their reliance on small, uniform transitions at each depth precludes dynamic, adaptive search capabilities. To address this, we introduce Bootstrap Flow-Map-Tree (a.k.a BFMT), a computationally efficient sampling framework that enables full tree-path construction from any tree depth using a single function evaluation, drastically reducing computational overhead while providing critical foresight for sequential sampling. By enabling dynamic transition time-step scheduling, BFMT efficiently allocates its sampling budget, smoothly transitioning from broad global exploration to fine-grained local refinement of high-utility modes discovered through exploration. Extensive experiments and ablation studies across diverse domains-from high-dimensional images to large-scale molecules-demonstrate BFMT's superiority over baseline approaches in both search and alignment tasks.

Large language models (LLMs) have demonstrated remarkable capabilities across diverse applications, yet ensuring their simultaneous safety, helpfulness, and trustworthiness remains a persistent challenge. Conventional refusal-oriented alignment strategies mitigate harmful content generation but systematically fail to serve legitimate user needs, often withholding information that could safely and constructively address the underlying intent of sensitive queries. Building upon the constructive safety paradigm pioneered by Oyster-I, which moves beyond blanket refusal toward thoughtful, response-oriented safety alignment, we identify two critical limitations of its Supervised Fine-Tuning (SFT)-based scheme: insufficient safety generalization to out-of-distribution scenarios and a phenomenon we term safety chain-of-thought (CoT) over-generalization, wherein safety-oriented reasoning patterns are excessively applied to benign queries, degrading helpfulness and user experience. To address these limitations, we propose Oyster-II, a reinforcement learning (RL)-based constructive safety alignment framework that adopts a Zero-RL paradigm combined with a multi-stage reinforcement learning strategy.Evaluated across extensive benchmarks, Oyster-II comprehensively surpasses both Qwen3-14B and its predecessor Oyster-I on safety dimensions, achieving cross-scale performance comparable to Qwen3-Max and Qwen3.5-397B.

LLM-based coding agents solve software-engineering tasks through iterative interactions with development environments, where returned observations accumulate in the context and become a major source of inference cost. Observation compression reduces this cost by shortening observations before they are appended to the context. However, existing methods still exhibit an unsatisfactory efficiency-effectiveness trade-off, as they do not explicitly model how compression affects the agent's subsequent behavior. This paper proposes CoACT, an action-preserving observation compression method for coding agents. CoACT is built on next-action preservation (NAP), which requires a compressed observation to induce the same next action as the raw observation. By checking the agent's immediate next action, NAP provides a practical signal for whether a compression preserves the information needed for continued task solving. During training, a teacher model first generates multiple compressed candidates of each observation. CoACT then uses an action-preservation reward based on NAP to filter out candidates that would change the agent's next action, and uses a length-reduction reward to choose compact candidates as supervision for a lightweight compressor. Experiments on SWE-bench Verified with three agentic models show that CoACT reduces average total token consumption by 33.0% while maintaining task-solving effectiveness close to the uncompressed agent.

Taxonomies provide key information about the semantic relationships between concepts and the inherent organization of vision and language. Despite their impressive capabilities, large multimodal models (LMMs) often lack taxonomic knowledge, leading to low hierarchical visual recognition (HVR) consistency. These models typically only rely on language modeling objectives during fine-tuning and lack explicit taxonomy-aware regularization. To address this, we propose Hierarchical Representation Regularization ($HiR^2$), a simple plug-and-play regularizer that improves hierarchical consistency in LMMs. Specifically, we introduce a semantic-aware visual tree construction framework that extracts coarse-to-fine visual features from intermediate LLM layers guided by textual cues. The regularizer combines two complementary objectives: a taxonomic entailment loss that enforces hierarchy via hyperbolic entailment cones in the Lorentz model, and a discriminative dispersive loss that promotes angular separation of semantically similar embeddings on the unit sphere without disturbing the radial hierarchical structure. Extensive experiments demonstrate that $HiR^2$ effectively captures taxonomic structures across diverse LMMs and fine-tuning methods. Code is available at https://github.com/PKU-ICST-MIPL/HiR2_ICML2026.

Explainability is central to building trustworthy AI, yet explanation interfaces can inadvertently provide adversaries with an expanded privacy-related attack surfaces. Recent studies show that advanced membership-inference attacks succeed by exploiting confidence-drop trajectories, induced through attribution-guided perturbations, as discriminative features, rather than directly using confidence scores or explanation vectors. Existing defenses against membership inference fail to directly mitigate such explanation-driven attacks. In this work, we investigate whether, during training, a model's own gradients can be leveraged as defense signals against such attacks, thereby aligning explanation profiles between members and non-members. To this end, we propose a Trajectory-Invariant Explanation Regularization (TIER) defense that penalizes erratic fluctuations in confidence drops simulated through gradient-guided perturbations and simultaneously minimizes the distributional shifts via KL-divergence. Unlike conventional adversarial training, which emphasizes label robustness, our approach targets explanation robustness by enforcing self-consistency through KL-divergence and reducing the variance of confidence drops between members and non-members. Extensive experiments confirm that our method effectively mitigates these attacks, delivering privacy protection while maintaining model utility and explanation fidelity.

Multimodal Large Language Models (MLLMs) have shown strong capabilities, but they may memorize private information from web data, raising privacy concerns. Machine unlearning offers a way to remove such private knowledge without retraining from scratch. However, existing MLLM unlearning benchmarks have two major limitations. First, they rely on simplified images that contain only the single target individual, failing to reflect the visual complexity of real-world photos. Second, they typically assume that the forget set and retain set are fully separated, ignoring the fact that private information is often visually entangled with benign public information. For example, a private individual may appear with a public figure or in front of a well-known landmark, where unlearning the private target should not damage the public context. To address these limitations, we propose PPE-Bench, a new benchmark for evaluating MLLM unlearning under private-public entanglement. Each image contains a target individual to be forgotten and public information to be preserved, including public figure and landmark. We further introduce two simple but effective methods to better preserve public information during unlearning. Through experiments, we find that existing unlearning methods can reduce private information leakage, but often substantially harm adjacent public information.

Deploying AI-generated video detectors in real-world services demands an ultra-low false positive rate (FPR) on real videos to avoid falsely rejecting authentic content, a regime where standard metrics such as AUROC fail to reflect actual operating behavior. We introduce Spatial Patch-Level Incoherence and Temporal Roughness (SPLIT), a training-free detector that operates on patch tokens from a frozen vision encoder to detect both fully generated and partially edited videos. SPLIT computes two complementary signals: Two-step Temporal Roughness (TTR), capturing non-smooth patch trajectories via one-step and two-step feature variation contrast, and Local Spatial Motion Incoherence (LSMI), measuring spatially inconsistent temporal changes through gradients of a feature-space motion field. The two are fused multiplicatively with gamma correction to sharpen real-fake separation at strict thresholds. We further propose a service-aligned evaluation protocol based on Fake Recall at fixed FPR with real-only threshold calibration and cross-real threshold transfer. Across three benchmarks (FakeParts, GenVideo, and ViF-Bench), SPLIT achieves the highest Fake Recall at FPR = $0.1\%$, substantially outperforming supervised and training-free baselines while remaining robust to post-processing with negligible overhead. The code is publicly available at https://github.com/mldljyh/SPLIT .

Cognitive Behavioral Therapy (CBT) provides a structured framework for understanding a user's mental state by examining the interaction between cognitive and behavioral factors. However, out-of-the-box LLMs respond fluently and empathetically, yet collapse into validation & reflection, regardless of what the user actually needs. They know theoretical CBT (scoring up to 96% accuracy on licensing exam questions) but fail to apply it effectively. We explore this gap with a knowledge-guided framework that treats CBT dialogue as controlled affective reasoning: user narratives are decomposed into Beck's Cognitive Conceptualization structure, grounded in clinical SNOMED CT concepts validated via Natural Language Inference, and a Multiple Chain-of-Thought (MCoT) strategy selection between Validation & Reflection, Socratic Questioning, or Alternative Perspectives. To measure whether such guidance actually changes behavior, we introduce the Protocol Leverage Force (F), a behavior-level metric that captures how far an intervention shifts a model away from its default response. Across three open-weight LLMs and 14 RealCBT-derived case studies, evaluated with human experts, valence-arousal trajectories, and linguistic entrainment, F shows that simply introducing protocol definitions via single chain-of-thought prompting fails to change LLM behavior, while MCoT on these definitions guides strategy selection better. Still, the effect stays within 1% (approx. 1.2-1.3%), and all models remain biased toward Validation & Reflection. These results show CBT knowledge alone does not ensure effective application, giving the affective-computing community instrumentation to measure where LLMs fall short.

Current benchmarks for evaluating large language models (LLMs) in medical calculation are largely based on simplified settings, where each patient case corresponds to a single calculator and the required tool is explicitly specified in the query. However, real clinical scenarios often require multiple calculators for joint evaluation, nested-scale calculation, and fuzzy queries that do not directly specify the target calculator. To this end, we propose a new medical calculation benchmark, MedCalc-Pro, which covers three progressively challenging task settings: single-calculator, multi-calculator, and nested-calculator calculation settings. MedCalc-Pro contains 2,268 real-world clinical cases, covering 77 medical calculators across 14 clinical departments. Meanwhile, to address the limited performance of existing frameworks and methods in complex clinical scenarios, we further propose a more generalizable agent framework that supports multi-tool selection and nested-tool calling, while suppressing parameter error propagation through structured validation and evidence review. We conduct systematic comparisons across open-source, closed-source, and medical-specialized LLMs, and the results show that our framework achieves the best performance across all three task settings. This work provides a new benchmark and method for evaluating and applying LLMs in challenging medical calculation scenarios.

Large language model agents driving security tool suites over the Model Context Protocol are increasingly common. Yet the factors that bound their capability remain poorly characterized: how much depends on the model versus the client that drives it, whether constraining the agent to the orchestrator's own tools helps, and where capability is limited by reasoning rather than by missing tools. Using HexStrikeAI, an open-source orchestrator that exposes 150+ tools, as a testbed, we follow a methodology that evaluates the system, diagnoses its failures, and applies targeted improvements. We run 86 picoCTF challenges across seven categories and three difficulty tiers, under three tool-access regimes and three model/client configurations (774 trials). We then apply corrections to existing tools, agent-behavior changes, and eleven new capability tools, and re-run the previously-unsuccessful trials. The diagnosis isolates the driving client as a first-order factor for a fixed model (a 2.1 * gap between two DeepSeek clients) and a monotonic difficulty gradient, with the largest gains in the mid tier. The overall solve rate rises from 55.4% to 72.0%, and every configuration improves significantly (paired McNemar p < 0.001, non-overlapping 95% confidence intervals). The residual failures are reasoning- or environment-bound rather than missing-tool. A 60-run stability sub-study finds single-run verdicts reproducible (17/20 unanimous). We discuss what the results imply for how such orchestrators should be evaluated, and we are explicit about the limits: the study uses a single benchmark, the fixes were tuned on the same challenges they were evaluated on, and the client effect is demonstrated for one model only, so its generality to other models remains a hypothesis.

Large language model (LLM) agents can improve through accumulated experience, but free-form textual memories become difficult to maintain, validate, and reuse as interactions grow. Recent symbolic approaches learn executable skills or programmatic world models, yet often store local procedures or assume simplified dynamics. We propose Object-Centric Environment Modeling (OCM), which organizes experience into an executable object-centric environment model. OCM maintains two connected code bases: object knowledge, which defines environment entities and mechanisms as Python classes, and procedure knowledge, which records reusable interaction patterns that must import and use the object model. OCM works in an online setting: after each episode, OCM reflects on the trajectory, updates both knowledge bases, and verifies that all procedures execute against the updated object model. During future interaction, the agent uses progressive knowledge disclosure to inspect compact code signatures first and read source code only when needed. Experiments show that OCM achieves the best average rank across benchmarks and reduces invalid actions, demonstrating that agents can benefit from building object-centric environment models.

Multi-view robotic manipulation methods with the attention mechanism have recently achieved significant progress in both training efficiency and task performance. However, the inherent redundancy, occlusion, and viewpoint dependency in robotic view images often lead to severe attention drift. To address this challenge, we propose AmpAttention, a novel attention mechanism inspired by differential amplifiers in analog circuits. It aims to suppress attention noise and capture high signal-to-noise ratio signals for more reliable perception. Based on this, we introduce the RVAF model, which integrates task-guided intra-view and inter-view AmpAttention. Compared to previous state-of-the-art methods, RVAF achieves the optimal average success rate across 18 RLBench tasks (249 variations) while reducing training time by 33.3%. RVAF also demonstrates strong potential in real-world high-precision tasks, exemplified by its ability to pick up a dart and accurately insert it into the red bullseye. Furthermore, we extend RVAF to RVAF++ by incorporating the SAM2 image encoder. RVAF++ achieves substantial gains on high-precision tasks, achieving a 91% success rate on the `insert peg' task. More qualitative results are provided at the anonymous project website https://anonymous.4open.science/w/RVAF-Anonymization.

Generative AI (genAI) systems produce cultural artefacts at scale, but they also reflect embedded cultural values through their design. Once identified, these values become open to deliberate reshaping. This position paper examines the maximalist values of current generative AI through an environmental humanities tradition and proposes design principles in which environmental sustainability serves as the core value instead. The principles are developed under the umbrella of Slow AI, a term that already circulates across several distinct research and practice programs. Five design principles are articulated (restraint, sufficiency, selectivity over retention, material visibility, and friction as affordance), each of them illustrated against the current design of widely deployed systems. Each principle operates at two levels: a design implementation, and an interpretive layer at which users and developers are prompted toward reflective engagement with the system. Together these principles extend human agency by restoring decisions that frictionless defaults have silently removed and do so by building interpretive reflection into design.

When a language model fails to answer a query on the first attempt, an agentic system retries, consuming additional tokens each time. This retry overhead creates a gap between what a model's per-token price implies and what a full workflow actually costs. We call this gap token inflation and define it as the ratio of true workflow cost to single-call cost. Systems like FrugalGPT route based on the latter, which can underestimate real cost by more than $2\times$ on difficult tasks. We address this with InflationAgent, a four-stage router that (1) measures token inflation systematically across model tiers and task types, finding inflation as high as $4.25\times$ for a 7B model on multi-hop question answering; (2) introduces CoT Branching Entropy (CBE), a pre-execution difficulty signal computed entirely from local inference, which predicts high inflation with AUROC 0.887; and (3) selects models by maximizing a Semantic Exchange Rate (SER) that divides expected accuracy by predicted true cost, with a fresh-escalation policy that discards failed chains before routing to a stronger model. On GSM8K under a fixed budget, InflationAgent achieves 94.7% accuracy versus 91.0% for FrugalGPT while using 31% fewer tokens, and we show that forwarding a failed reasoning chain to GPT-4o reduces its accuracy by up to 34.8 percentage points, validating the fresh-escalation design.

Generative artificial intelligence (GenAI) has challenged the validity of unsupervised online assessment, especially in technical subjects where plausible answers can be produced with little effort. This paper reports lessons from designing and implementing an AI-aware, AI-testing assessment in a large second-year undergraduate database systems module. The design combined two linked elements: (1) a structured three-part response format (X1-X2-X3) in which students documented a sourced answer, produced their own answer, and evaluated the sourced output; and (2) an AI-aware question-design process in which draft tasks were stress-tested against contemporary GenAI tools and revised when generic prompting produced superficially adequate answers. The account draws on archived assessment materials, rubrics, planning records, design-time GenAI trials, practice-response data, attainment records, and external review comments. Its main contribution is a reusable assessment-design method rather than a claim of measured learning gains. We show how the pattern developed across iterations and how it can support authentic assessment, visible AI literacy, student judgement, and more transparent marking. The paper offers practical guidance for lecturers adapting assessment to routine GenAI use, focusing on testing AI literacy rather than penalising students for misconduct.

The energy demand growth and environmental impacts of artificial intelligence (AI) have generated substantial interest in supplying sufficient low-cost electricity for AI-driven data center development. Research on the ability of demand-side management to address these challenges has been more limited. Shifting the amount or timing of demand from retail, corporate, and other organizational behaviors is a plausible option but only if changes in demand-related behavior have important effects on the envi- ronmental and electricity effects of AI. This article tests four retail (i.e., consumer) user behaviors with high behavioral plasticity to assess their technical abatement potential. The research concludes that non- reasoning models provide sufficient quality while consuming close to one-twentieth of energy compared to reasoning models, saving an amount equal to the annual electricity requirement of at least 141,000 US households under daily usage assumptions. Simple prompt modifications can yield additional reduc- tions in energy consumption by up to 65% using non-reasoning models. Specifically, the practice that maintains the highest degree of similarity with the baseline reduces electricity demand in the range of 4 to 35%, an amount equal to the annual electricity requirement of up to 7,200 US households. Although AI advancements make precise estimates of environmental and electricity impacts difficult to assess, the results confirm that certain minimally intrusive best practices aimed at the majority of users can reduce the energy and environmental burdens imposed by AI.

Longitudinal, in-the-wild, wearable sensing yields day-level physiology, sleep, activity, and environmental streams, whereas affect and cognition are labeled only episodically (per waves). We recast this cadence mismatch as a temporal representation problem and compare three wave-level mappings from dense histories to sparse labels: levels (within-wave summaries), absolute drift (change across waves), and proportional drift. Using almost a year of data from 82 adults in the Providemus alz study, we model 21 affect and cognition outcomes. Day-scale signals are reduced to compact wave-level descriptors (central tendency, dispersion, and distributional shape) and learned with four regressors under two orthogonal evaluation axes: leave-one-subject-out and leave-one-wave-out. Performance is reported as scaled MAE using both mean and median across folds. Differences emerge: affective states are best predicted by wave-to-wave absolute drift, whereas cognitive performance aligns with within-wave levels, reflecting emotion dynamic theories. Across windowing features, shape descriptors (e.g., minima, kurtosis) carry more signal than simple means/medians. We contribute a representation triad for sparse-label modelling, a wave-level feature schema applicable on-device, and a dual-axis reporting practice that separates cross-participant generalization from temporal robustness. These results convert temporal representation from an implicit preprocessing step into an explicit, testable design choice for real-world affective-computing applications in brain health.

In the literate human brain, reading and writing doubly dissociate: a ventral decoding route (pure alexia) and a fronto-parietal encoding route (pure agraphia), sharing a partial orthographic core. A decoder-only large language model (LLM) drives both from one autoregressive path optimized on text (a cultural invention, not an evolved instinct). We ask how entangled it is, comparing an input-side “reading code” $\mathbf{W}_{E}$ with an output-side “writing code” $\mathbf{W}_{U}$ via an index $\mathcal{E}\in[0,1]$ (CKA, Procrustes residual, mutual $k$-NN) calibrated against an independent-init floor and tied ceiling. On GPT-2, OPT and Pythia (14M–1.4B), untied models hold one coupled but sub-ceiling code ($\mathcal{E}=0.23$–$0.35$, far above floor) on a non-monotonic couple-then-differentiate trajectory, $\mathbf{W}_{U}$ drifting $\sim$3.2$\times$ farther than $\mathbf{W}_{E}$ in every decile. Equally informative is a negative: the matching behavioural test, that comprehension and production fail together rather than dissociate, cannot be run. For minimal pairs the alexia analogue is empty by theorem: greedy production implies a vocabulary-wide argmax, so it wins the pairwise ranking. Differential-damage indices are not scale-identified: heavy-tailed damage makes linear standardizations collapse onto their larger term, and the rank transform fixing this is bounded, so its null saturates. Both scores also contain the target's log-probability, which alone explains most of their variance and manufactures the apparent coupling. We withdraw a coupling statistic, a cross-level bridge and a separation measure. In a model reading and writing off one next-token distribution, no output-side pair isolates either ability: entanglement needing no index to see. By analogy, not homology, this situates LLMs in the space of possible minds.

GraphRAG enables deeper reasoning by structuring knowledge as graphs but struggles with n-ary facts. HyperGraphRAG uses hypergraphs for richer semantics, improving accuracy, yet relies on error-prone LLM extraction and inefficient standard chunk retrieval. We address this by employing self-consistency prompting to improve the extraction, and Personalized PageRank algorithm over hypergraph to enhance chunk retrieval.

While auditing a perturbation-based OOD detector on a document benchmark, we recorded an AUROC of 0.326 – well below the 0.5 chance level. The cause is a benchmark leak: the designated "OOD" class is one the model was trained on, so its examples sit inside the in-distribution fit set and the detector is penalized for correctly ranking them as familiar. Deleting the class and retraining 35 models across two domains raises the score to 0.911. We distill the contamination into a leak fingerprint – near-perfect supervised decodability (AUROC approximately 1) coupled with unsupervised detection collapsed below 0.65 – and validate it on a controlled battery of 52 settings (20 leaked, 32 clean) across ResNet-50 and ViT-B/16 on CIFAR-10/100, achieving sensitivity 18/20 and specificity 31/32 in embedding space; the matched fit-set-exclusion controls are perfect at 20/20. An in-the-wild audit of 24 standard near/far OOD benchmark pairs fires on exactly one (the intrinsically hard CIFAR-100 vs CIFAR-10 pair) and on no far-OOD pair, confirming specificity and that standard cross-dataset construction is clean. Under the corrected protocol, perturbation signals are decodable but not detectable: a supervised reader recovers the OOD signal (AUROC 0.87-1.00) while no unsupervised detector does, and the perturbation method does not improve on plain Mahalanobis distance. We provide a theoretical account of why and, for transparency, retract an earlier circular correlation. The contributions are a corrected protocol and a validated leak diagnostic, not a new OOD method.

Low-Rank Adaptation is widely used for parameter-efficient fine-tuning, yet existing methods typically assign the same adapter rank to every transformer layer despite their heterogeneous adaptation requirements. In this work, we show theoretically and empirically that uniform rank allocation is fundamentally suboptimal. Motivated by this observation, we propose LAARA (Layer Aware Adaptive Rank Allocation framework), a search-free framework that dynamically allocates ranks using lightweight diagonal Fisher estimates computed during training. LAARA combines projection-wise normalization, logarithmic compression, blended adapter importance estimation, and a vote-to-change dampening mechanism to produce stable and efficient rank adaptation. Experiments on GLUE and MathInstruct benchmark demonstrate that LAARA consistently matches or outperforms popular state of the art approaches such as LoRA, AdaLoRA, DyLoRA, and Bitfit while using significantly fewer trainable parameters. Our results show that Fisher-guided rank allocation provides a principled and effective foundation for adaptive parameter-efficient fine-tuning. The code is publicly available at: https://anonymous.4open.science/r/LAARA-D305/LAARA.py

As AI-driven Decision Makers (ADMs) influence our socioeconomic reality, their roles in both enhancing efficiency and amplifying the social biases have drawn significant attention. In this paper, we revisit the nuances of long-term fairness achievable by an ADM– specifically, through the lens of a loan approver inducing a population-level wealth dynamics. The literature generally (a) considers passive environments, i.e. the decisions of an ADM does not change the population's behaviour, and (b) measures bias in terms of disparity in the instantaneous decisions rather than downstream equity. Modern ADMs challenge both the notions. To address these caveats, we first formalise the wealth dynamics induced by a loan approving ADM interacting with a multi-demographic population as a Performative Markov Decision Process. Then, we mitigate the absence of such a performative test-bed by developing Eutopia: a lending-process simulator enabled with a novel performative data generator to learn long-term fair strategies. With Eutopia, we test reinforcement learning algorithms with different fairness-aware utilities dependent on approval decisions and downstream wealth. Results show that performative dynamics-aware learning with fairness-aware utility that incorporates the downstream outcomes induce better long-term equity and inclusivity.

Surrogate modeling for high-dimensional nonlinear dynamical systems that exhibit chaos requires mechanisms that preserve not only pointwise accuracy but also the scale-dependent structure of physical fields. Bandwise spectral power losses, such as the binned spectral loss function, provide such supervision on structured grids, where Fourier modes define a standard frequency decomposition. On irregular meshes, however, no canonical Fourier basis exists, and spectral representations must be constructed from graph operators induced by mesh connectivity and geometry. In this study, we extend the binned spectral power loss for application to unstructured-mesh surrogate modeling of nonlinear dynamical systems. This is obtained by replacing Fourier bands with graph-Laplacian frequency bands, and we provide scalable Chebyshev and multilevel approximations for improving long-horizon rollout fidelity. In its full-spectrum form, our approach uses graph Laplacian eigenspaces to provide a graph analogue of Fourier band-power matching, but incurs the high cost of spectral decomposition. As a scalable approximation, we replace exact band projectors with sparse Chebyshev polynomial graph filters, avoiding explicit eigendecomposition. When utilizing multilevel graph architectures, we introduce Graph Laplacian Energy Alignment for Meshes (GLEAM), which applies retained-subspace scale-aware supervision across graph hierarchies so that coarse and fine representations are regularized during autoregressive rollout. Our results show that the proposed spectral losses improve long-horizon rollout fidelity and preserve statistical invariants for the forecasting of turbulent flows on unstructured meshes, compared to deterministic baselines.

Baghchal is a two-player asymmetric board game with Nepali origins where four tigers are to capture goats and twenty goats desire to keep tigers in immobility. Although Baghchal has a complex structure which is strategic, has perfect information structure, and has cultural meaning, it has not been adequately covered in deep reinforcement learning (RL) literature. This paper gives a systematic exploration of four deep RL solutions Deep Q-Network (DQN), REINFORCE, Proximal Policy Optimization (PPO) and MuZero that are trained on one side of the asymmetric gameplay of Baghchal and then evaluated on the other side. The algorithms are rated based on win rate, draw rate, average captures, training convergence and computational cost. It is experimentally found that MuZero generates the best performance in both tasks, achieving 86 percent win over these Tiger and 62 percent win over these Goat and the ability to do so is due to the model-based planning machine through the Monte Carlo Tree Search. PPO is the most realistic algorithm and is provided to be competitive over both asymmetric tasks with significantly reduced computational costs compared to MuZero. Emergent strategic behavior analysis shows that model-based strategies are optimal over long-horizon planning, whereas value-based counterparts like DQN are more biased up towards the Tiger role owing to the more substantial reward signal.

A conventional codec stores a video as compressed pixel data. We instead store the video, together with its audio track, as the weights of a single sinusoidal representation network (SIREN) that maps space-time coordinates to RGB values and audio amplitudes. The network uses separate audio and video initialization layers, a stack of shared fully connected hidden layers, and three output branches: one for video and two Siamese audio branches whose disagreement is used to estimate and subtract residual noise. The overfitted teacher network is then compressed by response-based knowledge distillation into a smaller student, followed by 16-bit symmetric weight quantization and lossless LZMA2 (xz) encoding. On a 6.08 MiB test video, the quantized student reaches a video PSNR of 28.72 dB with SSIM of 0.75, and an audio PSNR of 24.18 dB with a log spectral distance of 10.69 dB, while the pipeline shrinks the representation from 9.05 MiB to 2.33 MiB, an overall compression ratio of 2.61. A bit-width sweep from 1-bit to 32-bit quantization shows that reconstruction quality saturates at 16 bits. We compare against H.264, HEVC, and MP3, report where the approach falls short of them, and describe a browser-based prototype that trains, transfers, and decodes these models over WebRTC.

Safe motion planning in advanced driver-assistance systems and autonomous vehicles requires an accurate understanding of how the surrounding traffic scene is likely to evolve. However, many existing lane-change prediction methods remain centered on a single target vehicle, while multi-agent forecasting approaches often describe scene evolution only through future positions and provide limited explicit information about the maneuver associated with each vehicle. This study proposes a dynamic scene graph attention framework that predicts the lane-change intention and future trajectory of every relevant vehicle within a local traffic scene. The scene is represented as a time-varying interaction graph in which vehicles are modeled as nodes and their spatial and kinematic relationships are encoded through explicit edge features. Temporal graph-attention message passing captures evolving inter-vehicle dependencies and pre-maneuver cues, while an intention-guided decoder links each predicted maneuver to its corresponding future motion. A scene-level consistency objective further encourages compatible multi-vehicle futures. Experiments on the NGSIM I-80, NGSIM US-101, and highD datasets demonstrate consistent improvements over competing baselines. DSiGAT achieves intention prediction accuracies of 90.12% and 90.97% on NGSIM I-80 and US-101, respectively, and reduces trajectory RMSE by up to 52.94% relative to the strongest baseline. It also produces lower inter-agent collision rates and joint displacement errors, indicating more coherent scene-level predictions. Ablation, sensitivity, robustness, and qualitative analyses further validate the contribution of the proposed components and the effectiveness of the scene-focused formulation.