arXiv:2605.30470v1 Announce Type: new
Abstract: Graph Machine Learning as a Service (GMLaaS) platforms increasingly implement explainability interfaces to meet regulatory transparency requirements. However, this transparency creates exploitable vulnerabilities for model extraction attacks. We present the first model extraction attack specifically designed for graph classification under strict black-box constraints where the attacker observes only discrete class labels and binary explanation masks (no probability scores, gradients, or confidence values). Our method (1) uses model explanation outputs to guide Monte Carlo edge sensitivity estimation toward decision boundaries, with Hoeffding concentration guarantees on estimation accuracy and (2) exploits explanation subgraphs to efficiently narrow the boundary search space. Extensive experiments on benchmark graph datasets across multiple domains demonstrate our method's superiority over comparable baselines. These findings demonstrate that such explainability interfaces create exploitable attack surfaces, informing both defensive mechanisms and policy frameworks for explainable AI mandates. The implementation code is provided in https://github.com/LabRAI/XSTEAL/.
Science Journals
arXiv:2605.28646v2 Announce Type: replace
Abstract: GUI agents rely on screenshots to infer intent and operate across applications, but these screenshots often contain private messages, medical records, payment credentials, and workplace-specific workflows. Privacy decisions in this setting depend on task, recipient, application state, and user role, yet static PII detectors miss these boundaries and cloud-side VLM reasoning can upload the raw screen before deciding what should be protected. We present MaskClaw, an edge-side privacy arbitrator for GUI agents. MaskClaw extracts local visual evidence, retrieves user- and task-specific policy memory, and decides Allow, Mask, or Ask before raw screenshots leave a trusted user- or organization-controlled environment. In five designed skill-evolution scenarios, it turns corrections, cancellations, and edits into reusable privacy skills checked by a sandbox gate. We introduce P-GUI-Evo, a benchmark built from real UI patterns, reconstructed HTML screens, and sanitized labels. Experiments show that pattern matching, cloud reasoning, and routing alone tend to over-confirm, over-mask, or expose raw screenshots under the same protocol. The artifact is available at https://github.com/Theodora-Y/MaskClaw.
arXiv:2510.03415v3 Announce Type: replace
Abstract: Recent work asks whether large language models (LLMs) condition their reasoning on explicit rules rather than statistical regularities from pretraining. Program execution provides a canonical instance: formal semantics define behavior through symbolic transition rules that can be systematically altered under distribution shift. We investigate whether LLMs can condition their reasoning on formal semantics through program execution and introduce PLSemanticsBench, pairing featherweight C programs with two semantic systems -- small-step operational semantics and K semantics -- and probing four capabilities: composing rules for final states, selecting rules when state is unmutated, sustaining such conditioning over long traces, and following supplied rules under novel semantics. To decouple semantic reasoning from syntactic familiarity, we redefine familiar operators to induce symbol-meaning conflict and introduce novel symbols defined only through the supplied rules, and stress-test models on Human-Written, LLM-Translated, and Fuzzer-Generated splits with increasing structural complexity. Across 11 frontier LLMs, strong final-state accuracy under standard semantics (up to 90%) drops sharply -- by as much as 40--60% points -- under semantic mutations and increasing structural complexity. Only a handful of models achieve non-zero long-horizon conditioning accuracy, and even the best systems reach just 35%. Together, these results suggest that contemporary LLMs often rely on pretrained lexical associations rather than systematically conditioning on supplied formal rules. PLSemanticsBench is publicly available at https://EngineeringSoftware.github.io/PLSemanticsBench.
arXiv:2605.30524v1 Announce Type: new
Abstract: Large language models are now adapted through chains of post-training stages rather than through a single instruction-tuning pass. This paper studies whether such sequential post-training gradually compresses internal representations into low-rank, anisotropic, and homogeneous feature spaces. We define a measurement suite for hidden states, logits, token trajectories, and LoRA updates, and we use it to analyze supervised fine-tuning, preference optimization, safety/refusal tuning, math and code specialization, and long chain-of-thought tuning under controlled stage orderings. The central hypothesis is that excessive representation concentration is not merely a geometric curiosity: it predicts reduced plasticity during later adaptation, weaker out-of-domain generalization, and poorer calibration. We further evaluate lightweight interventions, including mixed-domain replay, feature refresh, representation diversity regularization, and LoRA update decorrelation, as ways to preserve future learnability without giving up the behavioral gains of post-training.
arXiv:2512.00349v3 Announce Type: replace
Abstract: Are frontier AI systems becoming more capable? Certainly. Yet such progress is not an unalloyed blessing but rather a Trojan horse: behind their performance leaps lie more insidious and destructive safety risks, namely deception. Unlike hallucination, which arises from insufficient capability and leads to mistakes, deception represents a deeper threat in which models deliberately mislead users through complex reasoning and insincere responses. As system capabilities advance, deceptive behaviours have spread from textual to multimodal settings, amplifying their potential harm. First and foremost, how can we monitor these covert multimodal deceptive behaviors? Nevertheless, current research remains almost entirely confined to text, leaving the deceptive risks of multimodal large language models unexplored. In this work, we systematically reveal and quantify multimodal deception risks, introducing MM-DeceptionBench, the first benchmark explicitly designed to evaluate multimodal deception. Covering six categories of deception, MM-DeceptionBench characterizes how models strategically manipulate and mislead through combined visual and textual modalities. On the other hand, multimodal deception evaluation is almost a blind spot in existing methods. Its stealth, compounded by visual-semantic ambiguity and the complexity of cross-modal reasoning, renders action monitoring and chain-of-thought monitoring largely ineffective. To tackle this challenge, we propose debate with images, a novel multi-agent debate monitor framework. By compelling models to ground their claims in visual evidence, this method substantially improves the detectability of deceptive strategies. Experiments show that it consistently increases agreement with human judgements across all tested models, boosting Cohen's kappa by 1.5x and accuracy by 1.25x on GPT-4o.
arXiv:2605.30556v1 Announce Type: new
Abstract: Random, untrained neural networks consistently match or exceed trained networks in representational similarity to early visual cortex. This puzzling finding challenges the assumption that learning improves brain alignment. We investigate it by tracking representational similarity analysis (RSA) alignment to human fMRI data across training for four learning rules: backpropagation (BP), feedback alignment (FA), predictive coding (PC), and spike-timing-dependent plasticity (STDP). Using 720 object images from the THINGS database and fMRI data from three subjects across six visual ROIs, we measure Spearman correlations between model and brain representational dissimilarity matrices at eight training checkpoints (epochs 0-40). We find that (1) a single epoch of training reduces V1 alignment by 25-90%, depending on the learning rule; (2) backpropagation reduces V1 alignment most severely (delta r = -0.080), while predictive coding and STDP preserve substantially more (delta r ~ -0.04); and (3) a weaker, opposite tendency appears in object-selective cortex (LOC), where BP shows the largest increase in alignment during training, although the absolute change is small. These results suggest that untrained architectures capture low-level visual statistics through inductive biases alone, and that global error signals (BP) reshape early representations more aggressively than local learning rules (PC, STDP), which better preserve brain-like structure.
arXiv:2605.31313v1 Announce Type: new
Abstract: In this work, a systematic study of electromagnetic instabilities driven by the temperature gradient in magnetically confined fusion plasmas with reversed magnetic shear is conducted using gyrokinetic particle-in-cell simulations. An electromagnetic instability arising in the low-beta regime is investigated, where beta=8*pi*nT/B^2 denotes the ratio of plasma pressure to magnetic pressure. Within a reversed shear safety factor (q) profile, when a mode rational surface coincides with the position of zero shear, an instability dominated by only one poloidal harmonic emerges, rather than the conventional ion-temperature-gradient (ITG) mode. Simulation results demonstrate that the instability exhibits pronounced electromagnetic polarization even in the low-beta regime, with a real frequency significantly higher than that of ITG modes, and show that it is destabilized by the temperature gradient and not by the density gradient. This instability can be observed even for a monotonic q profile with weak magnetic shear. Based on a systematic comparison with other typical electrostatic and electromagnetic instabilities, this instability is identified as a weak shear Alfvenic-ion-temperature-gradient (WSAITG) mode, which may provide an explanation for the low-frequency Alfven modes (LFAM) observed in experiments. Wave-particle resonance analysis in phase space reveals that, in contrast to the ITG mode, well-passing particles provide an additional resonant population that drives the WSAITG mode.
arXiv:2605.30709v1 Announce Type: new
Abstract: A detailed understanding of the capabilities and fidelity of low-energy reconstruction is crucial for taking advantage of MeV-scale neutrino physics opportunities in liquid argon time projection chambers (LArTPCs). This study presents a measurement of the resolution of reconstructed energy in the MicroBooNE LArTPC at $\approx 1.5$ MeV. The characterization is performed using monoenergetic signals generated by $2.614$ MeV $\gamma$-rays from $^{208}$Tl decays undergoing pair production in the detector. The resolution is found to be ($7.52 \pm 0.78 \text{(stat)} \pm 0.92 \text{(syst)}$)%. This value is consistent with the MicroBooNE simulation prediction of ($9.70 \pm 0.65 \text{(stat)}$)% at the $1.6 \sigma$ level. This study represents the first ever measurement of LArTPC energy resolution at the MeV scale and provides a pathway for monoenergetic energy calibrations in future experiments using LArTPC detectors.
arXiv:2605.30695v1 Announce Type: new
Abstract: Deploying vision-language-action (VLA) policies in industrial environments requires the ability to teach new tasks at low cost, a property current VLAs lack, since each new task requires fine-tuning. We investigate whether primitive-aware training produces a transferable artifact: a learned library of sub-skills that can be composed at inference time, conditioned on a small number of demonstrations, to perform tasks the policy was never trained on. We train two VLA architectures with different inductive biases, OpenVLA and $\pi_{0.5}$, on the REASSEMBLE contact-rich assembly dataset under matched LoRA fine-tuning recipes and locked hyperparameters, varying training between flat trajectories and primitive-segmented episodes with primitive-specific language prompts. We hold out 6 object-task combinations from training and evaluate few-shot transfer: models receive $m \in \{0, 1, 3, 5, 10\}$ demonstrations of a held-out task and attempt execution without weight updates. We replicate across three training seeds and validate on a second dataset (LIBERO-Long). Primitive-trained models reach 78% of fine-tuned upper-bound performance with only m=3 demonstrations, while flat-trained models require m=10 demonstrations to reach the same level -- a $3\times$ sample efficiency gap that replicates across seeds, architectures, and datasets. To establish causation, we ablate the primitive-decodable subspace of hidden states and show few-shot transfer degrades by 32 percentage points while ablating a random subspace of equal dimensionality has no effect, indicating primitive representations are causally necessary rather than incidentally correlated with transfer. We identify and correct a methodological pitfall in evaluating chunked policies: family-wise inflation of single-step action-range gates produces order-of-magnitude higher false-failure rates against ground-truth human demonstrations.
arXiv:2605.31146v1 Announce Type: new
Abstract: This paper investigates how User Experience Research (UXR) methods can be combined with AI-supported analysis to develop clearer design direction for digital wellbeing interventions targeting Emergency and Public Safety Personnel (EPSP). EPSP work in high-stress, shift-based environments where cognitive fatigue and unpredictable schedules reduce engagement with conventional wellbeing tools. Using the UXR Point-of-View (PoV) framework, this study applied an AI-supported literature analysis process to identify recurring psychological, behavioural, and design patterns. Behaviour Change Techniques and Persuasive Technology principles were integrated throughout interpretation to connect evidence with practical design reasoning. The process resulted in a UXR PoV Pyramid, nine UXR Play Cards, and stakeholder focused PoV narratives. Findings show that effective wellbeing systems for EPSP must minimise cognitive effort, adapt to operational context, and prioritise psychological safety. The work demonstrates how AI can assist large-scale evidence interpretation while human researchers maintain responsibility for contextual judgement and design direction.
arXiv:2605.31149v1 Announce Type: new
Abstract: This study investigates how UX research (UXR) principles, combined with Large Language Model (LLM)-supported analysis, can be used to improve the quality of requirements for mobile learning systems designed for learners with cognitive disabilities. Using the UXR Point-of-View (PoV) pyramid as a methodological framework, the study progressed through four stages: foundational structuring of psychological, behavioral, and design layers; structured validation using the DeLone and McLean Information Systems Success Model and Quality Function Deployment (QFD); insight consolidation through the development of nine Cognitive Accessibility UXR Play Cards; and stakeholder-specific PoV articulation to support interdisciplinary communication. LLM-supported synthesis was integrated to assist in theme clustering, requirement refinement, and hypothesis formulation under human oversight. Findings suggest that many usability and engagement challenges in mobile learning originate from ambiguous or under-specified requirements rather than interface design alone. By embedding cognitive accessibility principles into measurable and technically traceable requirements, the proposed Cognitive Accessibility UXR Playbook provides a structured pathway for aligning theory, system architecture, and stakeholder strategy.
arXiv:2605.27996v2 Announce Type: replace
Abstract: Single-axis mitigations of reward-model biases (e.g., reducing proxy reliance on length, sycophancy, or style) can rotate optimization pressure onto correlated proxies rather than eliminate it, a failure mode we call reward bias substitution. The failure is enabled by a measurement-versus-optimization gap between audit and policy-induced distributions during mitigation evaluation and policy training. We formalize mitigation outcomes into a regime taxonomy and prove that successful mitigation, bias substitution, and overcorrection produce identical observables under any audit-distribution scoring, including ranking accuracy and win-rate, even when granted oracle access to the true reward. Across published preference-learning mitigation work, no method we survey reports the evidence needed to certify successful mitigation. Augmenting evaluation with policy-induced distributions while tracking multiple biases provably closes the gap, and we translate this into actionable prescriptions for mitigation methods and benchmarks. We demonstrate bias substitution in language model RLHF, where a length penalty during GRPO training compresses responses as intended yet redirects optimization pressure onto confidence calibration, driving the policy into overconfidence while factual free-form accuracy falls. We also show a published length-debiasing operator that zeroes reward-length correlation on the audit distribution but reintroduces bias under best-of-N selection on three of four SOTA reward models, and a length-sycophancy coupling whose direction reverses under human-LLM judge disagreement.
arXiv:2603.30015v2 Announce Type: replace-cross
Abstract: Interactive verification protocols for quantum computations allow to build trust between a client and a service provider, ensuring the former that the instructed computation was carried out faithfully. They come in two variants, one without quantum communication that requires large overhead on the server side to coherently implement quantum-resistant cryptographic primitives, and one with quantum communication but with repetition as the only overhead on the service provider's side. Given the limited number of available qubits on current machines, only quantum communication-based protocols have yielded proof of concepts.
In this work, we show that the repetition overhead of protocols with quantum communication can be further mitigated if one examines the task of operating a quantum machine from the service provider's point of view. Indeed, we show that the test rounds data, whose collection is necessary to provide security, can indeed be recycled to perform continuous monitoring of noise model parameters for the service provider. This exemplifies the versatility of these protocols, whose template can serve multiple purposes and increases the interest in considering their early integration into development roadmaps of quantum machines.
arXiv:2603.22867v1 Announce Type: cross
Abstract: Multimodal stacks that mix ViTs, CNNs, GNNs, and transformer NLP strain embedded platforms because their compute/memory patterns diverge and hard real-time targets leave little slack. TRINE is a single-bitstream FPGA accelerator and compiler that executes end-to-end multimodal inference without reconfiguration. Layers are unified as DDMM/SDDMM/SpMM and mapped to a mode-switchable engine that toggles at runtime among weight/output-stationary systolic, 1xCS SIMD, and a routable adder tree (RADT) on a shared PE array. A width-matched, two-stage top-k unit enables in-stream token pruning, while dependency-aware layer offloading (DALO) overlaps independent kernels across reconfigurable processing units to sustain utilization. Evaluated on Alveo U50 and ZCU104, TRINE reduces latency by up to 22.57x vs. RTX 4090 and 6.86x vs. Jetson Orin Nano at 20-21 W; token pruning alone yields up to 7.8x on ViT-heavy pipelines, and DALO contributes up to 79% throughput improvement. With int8 quantization, accuracy drops remain <2.5% across representative tasks, delivering state-of-the-art latency and energy efficiency for unified vision, language, and graph workloads-in one bitstream.
arXiv:2605.31512v1 Announce Type: new
Abstract: Multilingual orthopedic decision support remains challenging in low-resource healthcare settings, where clinical narratives contain specialized terminology, mixed scripts, incomplete evidence, label imbalance and language-dependent documentation patterns. This article presents a reliability-oriented framework for classifying free-text orthopedic notes in English, Hindi and Punjabi. We compare task-aligned multilingual transformer encoders, a task-fine-tuned DistilBERT baseline, zero-shot instruction-tuned large language models (LLMs) and a domain-adaptive encoder, IndicBERT-HPA. IndicBERT-HPA augments IndicBERT with language-aware orthopedic adapter heads to support clinically relevant multilingual representation learning. Evaluation extends beyond aggregate accuracy to per-class performance, ROC-AUC, AUPRC, expected calibration error, cross-language stability and robustness under controlled balanced and natural-prevalence distributions. The evaluated zero-shot LLMs remain substantially less effective than task-adapted encoders for closed-set classification, with language-dependent instability. Under natural clinical prevalence, IndicBERT-HPA achieves the strongest overall performance, reaching an averaged Macro-F1 of 0.8792, Macro-AUROC of 0.894 and AUPRC of 0.902. We further implement a deterministic selective-verification layer combining confidence gating, evidence-consistency checking and language-risk screening. On a randomly selected held-out 5,000-record subset, it achieves 84.4% selective accuracy and 0.76 selective Macro-F1 at 72.3% coverage, compared with 71.5% accuracy and 0.65 Macro-F1 for accept-all prediction. These results support reliability-oriented multilingual clinical decision support with explicit deferral.
arXiv:2605.25134v3 Announce Type: replace
Abstract: Sparse optimization is a fundamental challenge in various practical applications. A popular approach to sparse optimization is $\ell_p$ regularization. However, it may encounter optimization instability due to the unbounded gradients when $0<p<1$. In this paper, we introduce a novel approach to sparse optimization termed ReWA, based on Reparameterization, Weight decay, and Adaptive learning rate. ReWA is closely connected to $\ell_p$-regularization, yet it unveils a distinct optimization landscape that helps mitigate instability issues. Experiments on CIFAR-10 and ImageNet with ResNets demonstrate that ReWA leads to significant sparsity improvements over the $\ell_1$-regularization approach while preserving test accuracy.
arXiv:2503.21168v3 Announce Type: replace
Abstract: Robots navigating human-populated environments must avoid collisions while respecting the social structure of crowds, particularly the implicit boundaries of social groups. Most navigation approaches model humans as independent individuals,causing socially disruptive behavior even when collision-free. This paper presents TAGA (Tangent Action for Group Avoidance), detected group boundaries via tangent-path maneuvers without modifying the underlying navigation policy. A hierarchical safety controller coordinates group-level avoidance with individual collision prevention. We propose the Group Crossing Rate (GCR), a continuous metric measuring the fraction of timesteps the robot spends inside any group convex hull, providing finer-grained social compliance assessment than terminal metrics alone. We introduce a realistic crowd simulation benchmark with five empirically grounded phases: individual speed heterogeneity, group speed coupling, F-formation static groups, leader-follower dynamics, and convex-hull boundaries, evaluated under both ORCA and Social Force pedestrian dynamics. Experiments across ORCA, Social Force, DS-RNN, and Intention-RL reveal a reactive-learning asymmetry: TAGA provides the largest gains for classical reactive baselines (up to +8pp success rate, GCR halved) with near-zero cost for learned policies. These findings offer actionable guidance for when modular group-awareness adds value versus when end-to-end group-aware training is preferable.
arXiv:2605.28952v2 Announce Type: replace
Abstract: E-values have attracted considerable interest in recent years as flexible tools for enabling anytime-valid and adaptive data analysis. Hypothesis testing is at the core of many of these applications, which can often involve private or sensitive data. In this work, we answer a simple but important question: given two distributions $\mathbb{P}$ and $\mathbb{Q}$, what is the maximum achievable e-power when testing $X\sim \mathbb{P}^n$ against $X\sim\mathbb{Q}^n$ with e-values that satisfy $\varepsilon$-differential privacy? We characterize the optimal rate for this problem and provide an algorithm which matches it exactly. In the sequential setting, when observations arrive one-by-one and the analyst chooses when to halt, we give matching upper and lower bounds on the stopping times of any private e-process. Numerical experiments confirm the practicality of our algorithms, which require less data than the recently proposed DP-SPRT across a range of sequential testing problems and privacy levels.
arXiv:2605.27557v2 Announce Type: replace
Abstract: Card payment fraud detection is usually framed as a supervised classification problem. Although this approach has generated practical progress, improvement has remained incremental despite major advances in model architecture. We argue that this is not mainly a failure of function approximation or optimization, but a consequence of structural information impairments inherent to the payment ecosystem.
We formalize card authorization as a sequential decision problem with delayed, censored, corrupted, and counterfactually missing feedback. We derive a minimax regret lower bound showing that these impairments enter multiplicatively in the denominator of the achievable learning rate. The bound implies that improving issuer reporting quality or reducing censorship can yield larger reductions in the regret floor than increasing model complexity. We also show that heterogeneity across issuers worsens learnability beyond what average impairment rates suggest.
The paper contributes a theory of why fraud detection in payment networks is fundamentally harder than in standard online learning settings, identifies ecosystem information quality as the key bottleneck, and provides a theoretical basis for prioritizing investments in reporting infrastructure, dispute process quality, and selective exploration. The paper is theory-first and does not rely on proprietary transaction data.
arXiv:2605.31340v1 Announce Type: new
Abstract: The appropriateness of empathy in AI has emerged as a critical concern, as excessive empathy risks seeming manipulative while insufficient empathy appears dismissive. While prior research has explored how to quantify empathy in AI, few studies examine whether such empathy is contextually appropriate. This paper introduces an economic perspective by applying signaling theory to human-AI conversations. We propose Signal Cost Proxies (emotional richness, perspective-taking, and contextual tailoring) mapped to affective, cognitive, and associative empathy. This multidimensional framework enables systematic evaluation of empathy not just by presence, but by its appropriateness relative to user demand.
arXiv:2605.30948v1 Announce Type: cross
Abstract: Higher-order interactions admit richer structural heterogeneity than pairwise networks. To understand how heterogeneity impacts collective phenomena we develop a framework based on the cavity method and apply it to the simplicial Ising model on heterogeneous hypergraphs. Unlike in homogeneous structures, group size and node degree play fundamentally different roles: size heterogeneity sharpens the transition via large-group unanimity, while degree heterogeneity softens it as hubs cooperatively seed ordering with non-hubs. Under either type of heterogeneity, continuous--discontinuous double transitions can arise, where the symmetry-breaking continuous transition is driven by pairs or by hubs, respectively. When both heterogeneities coexist, cross-order degree correlations further modulate the phase diagram, with anticorrelation delaying the group-driven discontinuous jump and broadening the hysteretic region. Our results reveal the intricate interplay between size and degree heterogeneities in collective phenomena beyond pairwise interactions.
arXiv:2605.31438v1 Announce Type: new
Abstract: Time series forecasting often requires learning nonlinear and time-delayed dependencies. A paradigmatic class of forecasting models are nonlinear vector autoregressive processes (NVAR), also known as next-generation reservoir computers (NG-RCs). These models approximate the Koopman operator on the space spanned by their explicit feature library. We consider the identifiability problem for learning Markovian nonlinear dynamical systems and show that the training error as a function of time resolution follows characteristic (pre-)asymptotic scaling laws. These laws depend on whether the feature library can represent the early Lie-series coefficients of the flow map (propagator) exactly or merely approximately. For dynamical systems governed by polynomial vector fields, we demonstrate the mechanism for NVAR/NG-RC models with monomial and Fourier feature libraries. We determine the dependence of the training error on the temporal resolution, the involved nonlinear degree, and the number of delay terms. While delay terms reduce the optimal one-step training error, they improve long-horizon forecasts only when the library provides sufficient nonlinearity. Thus, small training error coexists with weak generalization as the model class is mismatched to the true data-generating process. Numerical experiments on various chaotic dynamical systems confirm the theoretical predictions.
arXiv:2605.31404v1 Announce Type: new
Abstract: Large Language Model (LLM)-based navigation systems commonly construct explicit spatial representations (e.g., topological graphs, semantic raster maps) and translate them into textual descriptions as LLMs' inputs. However, the linguistic structures of such text-based spatial representations and the choices of contextual features (e.g., topology, geometry) they contain are often treated as neutral engineering decisions rather than key factors that shape LLMs' behavior. To fill the gap, we propose a dual-interventional framework that disentangles linguistic structures from different contextual cues to evaluate the linguistic inductive bias of LLMs for navigation planning. In the framework, representation intervention varies the linguistic format and the degree of linguistic compression, clarifying when linguistic representations support or inhibit navigation planning. Context intervention, combined with contextual feature combination and conflict probing, explicitly clarifies the preferences and weaknesses of LLMs when processing different contextual cues. Experiments across diverse spatial reasoning tasks and multiple model scales reveal a consistent pattern: topological information is a sturdy shield and the backbone of robust planning; linguistic format is a double-edged sword whose effect depends on model size, task demands, and the compression level; and semantic information is a fatal Achilles' heel -- incorrect semantic cues can systematically derail the planning process. Overall, our study shows that effective text-based spatial representations in LLM-based navigation should preserve topological integrity, calibrate representational compression to model capacity, and ensure semantic correctness, rather than simply adopting a single representation. Our code is publicly available at https://github.com/jonesdong150/LLM-Navigation-Inductive-Bias.
arXiv:2605.31392v1 Announce Type: new
Abstract: Polyconvex constitutive modeling is attractive as it guarantees stability of numerical simulations and can improve the generalization behavior of material models. However, in certain applications, polyconvex formulations perform poorly in reproducing the underlying ground truth material response, which can effectively preclude their practical use. In this work, we address this issue and investigate the limitations of polyconvex constitutive modeling. The main contributions of this paper are as follows: (1) We analyze the theoretical reasons why polyconvexity may, in some cases, impose overly restrictive constraints that limit the achievable accuracy of constitutive models. Thereby, we provide analytical ellipticity guarantees for two non-polyconvex Mooney-Rivlin type potentials. (2) We investigate the practical limitations of polyconvex physics-augmented neural network constitutive models using two representative formulations: models using structural tensor-based invariants and models using signed singular values. Their performance is evaluated on datasets obtained from homogenized microstructured materials, and their predictive capabilities are assessed in finite element simulations. (3) Overall, we provide an overview of benefits, limitations, and mitigation strategies of polyconvex constitutive modeling.
arXiv:2605.30907v1 Announce Type: new
Abstract: We present BlueFin, a benchmark that tasks large language model (LLM) agents with synthesis, manipulation, and comprehension tasks over spreadsheet workbooks in the professional finance domain. Though estimates of the global population of paying users of spreadsheet software range in the hundreds of millions -- an order of magnitude more than the estimated global population of professional developers -- comparatively fewer resources have been devoted to exploring and expanding LLM capabilities in the spreadsheet domain, with fewer still dedicated to mirroring real occupational tasks encountered by those in professional finance roles. In response, we curate a set of 131 challenging, complex tasks with real-world relevance in the domain, containing 3,225 granular rubric criteria; notably, our rubric criteria and LM judge evaluations are validated by a team of expert human annotators, resulting in high-quality, granular evaluations of complex tasks that are difficult to verify programmatically but can be reliably evaluated by an LM judge agent. Our judge achieves parity with expert consensus ($\alpha=0.826$) with a macro-F1 score of 0.839. Frontier LLMs demonstrate poor performance on the challenging benchmark, with the strongest LLMs achieving less than 50\% average scores across tasks -- models exhibit particular weaknesses in dynamic correctness. Our contributions include a dataset of examples across three categories of spreadsheet tasks, an open source harness and agentic evaluation framework, and a characterization of existing frontier models' performance on our benchmark.