Academic Intelligence · Curated Daily

Explore the Frontier of Global Academia

AcademicHub aggregates real-time literature from top journals and preprint platforms. Build your personal research radar and let large language models compile cross-disciplinary analysis briefings automatically.

01.
arXiv (CS.CL) 2026-06-16

LLM Judges Have Dark Current: A Psychometric Datasheet for LLM-as-a-Judge Evaluation

LLM-as-a-judge systems are now routinely used for open-ended model evaluation, where human preference annotation is costly, slow, and difficult to reproduce. Yet these judges are often reported as scalar accuracy, win-rate, or agreement devices. We argue that a judge should instead be reported as a measurement instrument. We introduce a Judge Datasheet protocol that measures dark current under true-vacuum inputs, stable cross-sensitivity to same-quality surface variation, positional false preference, target sensitivity on a controlled quality ladder, and the criterion or operating point induced by tie instructions. The direction-stability decomposition reveals that apparent Delta0 preference can be stable surface response or disguised position bias. In a three-judge open-weight case study, Llama-3.1-8B shows high dark current and presentation-conflicted Delta0 behavior, Qwen2.5-14B is vacuum-clean and target-sensitive but mixes stable and positional over-discrimination, and Qwen2.5-32B is vacuum-clean with low stable cross-sensitivity and low positional false preference. A strict tie criterion eliminates Qwen32B Delta0 false preference but absorbs marginal Delta1 target signals into ties while preserving Delta5 sensitivity. The results show that prompting moves the criterion, not the resolution. We do not claim that the downstream mechanism hypothesis that motivated this work is confirmed; the contribution is a metrological protocol for measuring the measuring device before downstream claims are made.

02.
arXiv (CS.CV) 2026-06-15

Avatar V: Scaling Video-Reference Avatar Video Generation

Generating avatar videos that are not merely visually similar to a target individual but behaviorally recognizable, faithfully reproducing their talking rhythm, gestural tendencies, and expression dynamics, remains an open challenge. Existing methods predominantly condition on single static images, which provide insufficient identity information and cannot capture dynamic motion traits, while standard pixel-level objectives underserve the perceptually critical facial regions that determine avatar fidelity. We present Avatar V, a production-scale framework that addresses these limitations through video-reference-conditioned identity modeling. Rather than compressing identity into fixed-size embeddings, the model conditions directly on the full token sequence of a reference video, learning to reproduce both static identity attributes (facial geometry, skin texture) and dynamic behavioral patterns (talking rhythm, micro-expressions) through attention over the reference context. We introduce Sparse Reference Attention, an asymmetric mechanism achieving linear-complexity conditioning on arbitrarily long references; a motion representation stream enabling closed-loop talking style transfer; and an identity-aware super-resolution refiner inheriting the full reference conditioning. These are supported by a data engine curating 100M+ training clips from 50M raw videos, and a five-stage training pipeline with flow matching pre-training, personality fine-tuning, two-phase distillation (>10x acceleration), and RLHF alignment, deployed across thousands of GPUs. Avatar V generates 1080p videos of unlimited duration, achieving state-of-the-art identity preservation, lip synchronization, and generation quality on our cross-scene benchmark, consistently outperforming leading systems including Seedance 2.0, Kling O3 Pro, Veo 3.1, and OmniHuman 1.5 in both automated metrics and human evaluation.

03.
medRxiv (Medicine) 2026-06-17

Treatment of Multi-Drug-Resistant Tuberculosis with Second-Line All-Oral Drugs in Ghana: Incidence of Adverse Events.

Introduction: The treatment of multidrug-resistant tuberculosis (MDR-TB) remains challenging due to the toxicity of second-line medications and suboptimal treatment outcomes. This study aimed to determine the incidence of adverse events and identify factors associated with these events in patients undergoing treatment for MDR-TB with second-line all-oral drugs in Ghana. Methods: This retrospective cohort study reviewed the medical records of 384 MDR-TB patients treated with second-line all-oral drugs at selected health facilities in Ghana, including the Greater Accra Regional Hospital, Eastern Regional Hospital, and Kumasi South Hospital. Data were extracted using the Kobo Collect tool, capturing patient demographics, baseline clinical and laboratory characteristics, treatment regimens, and adverse events. The study period spanned from 2020 to August 2024. Results: The study included a total of 384 MDR-TB patients, with a mean age of 45 years (SD = 15). The majority of patients were male (65.78%), and most were within the 45-64 years age group (33.85%), followed by those aged 25-44 years (31.25%). Regionally, the highest number of cases were reported from the Greater Accra Region (39.06%), followed by the Eastern Region (31.25%) and Kumasi South Hospital (29.69%). Approximately one in four patients (25%) presented with comorbidities, with HIV being the most common (19.5%). The most frequently reported adverse events were diarrhea (14%), dizziness (13.7%), and vomiting (12.3%). Most of these were mild to moderate in severity and tended to decrease as treatment progressed. Severe adverse events, such as leukopenia and acute kidney injury, were rare, occurring in less than 5% of patients. Over the course of treatment, gastrointestinal adverse events such as vomiting and nausea showed a significant decline, indicating possible patient adaptation or improved clinical management. Results from the multivariate Poisson regression analysis revealed that age and comorbidities were significant predictors of adverse events. Patients aged 65 years and above had a 56% lower risk of developing adverse events compared to younger patients (Adjusted Risk Ratio [aRR] = 0.44, 95% CI: 0.25-0.79, p = 0.005). Conversely, patients with comorbid conditions such as diabetes or hypertension were approximately 2.6 times more likely to experience adverse events compared to those without comorbidities (aRR = 2.65, 95% CI: 1.58-4.43, p < 0.001). The effect of sex was not statistically significant after adjustment (aRR = 1.03, 95% CI: 0.70-1.50, p = 0.86). At the end of the treatment period, 74.9% of patients achieved successful outcomes, including both those who were cured and those who completed treatment without being classified as cured. However, 25.1% had unsuccessful outcomes, which included treatment failure, relapse, or death. Conclusion: In conclusion, adverse events are common in the treatment of MDR-TB with second-line All-Oral drugs, with gastrointestinal adverse events being the most prevalent. These findings highlight the importance of monitoring and managing adverse events to optimize treatment outcomes for MDR-TB patients in Ghana.

04.
arXiv (CS.CL) 2026-06-18

RegMix-D: Dynamic Data Mixing via Proxy Training Trajectories

Data mixture selection is critical for Large Language Model pretraining. Existing methods such as RegMix select a single static mixture by fitting a regression model on small-scale proxy runs. We propose RegMix-D, a simple extension of RegMix to dynamic mixing. Our key observation is that proxy runs produce not only endpoint losses, but also full loss trajectories, which can be used to further improve data mixture. By training regression model on these trajectories, we can predict optimal mixtures at multiple training stages. RegMix-D supports two deployment modes: an offline variant that generates a complete mixture schedule before target training, and an online variant that adapts the mixture during training using observed loss. Experiments on 25B tokens of the Pile dataset with a 1B parameter target model show that RegMix-D consistently improves over RegMix and DoReMi across 13 downstream tasks while remaining proxy-efficient: it surpasses RegMix even with only 128 proxy models (25% of RegMix's proxy compute budget).

05.
arXiv (CS.LG) 2026-06-16

Generative Modeling on Metric Graphs via Neural Optimal Transport

arXiv:2606.16273v1 Announce Type: cross Abstract: We introduce, to our knowledge, the first deep generative modeling framework for probability distributions continuously supported on compact metric graphs. Given source and target measures on a metric graph, our method embeds the graph into a smooth ambient space, solves an entropic Kantorovich problem via a neural semidual parameterization, and projects generated samples back onto the original graph. We study two embedded geometries: an extrinsic Euclidean realization and the intrinsic tropical Abel–Jacobi embedding into the Jacobian torus. In both cases, the resulting generator is graph-supported by construction. We prove that, in the joint limit of increasing neural expressivity, the learned generator converges weakly to a valid transport coupling between the original graph measures. Empirically, across a range of geometrically distinct graphs, our method matches or improves upon heuristic transport baselines based on discrete graph OT, while scaling more favorably. Finally, we demonstrate scalability on real-world urban mobility data by training our model on one million Uber pickup locations in Manhattan, New York City.

06.
arXiv (CS.LG) 2026-06-17

Generalization Guarantees for Multi-Input Neural Operator Learning in Sobolev Spaces

arXiv:2606.17419v1 Announce Type: new Abstract: We develop approximation and generalization error estimates for multi-input neural operators, with the output error measured in Sobolev norms. In contrast to standard operator-learning settings with a single input function, our framework allows multiple input functions defined on possibly different domains, with different dimensions and Sobolev regularities. The derived rates explicitly quantify the contribution of each input space to the final error bound. In particular, in the balanced regime, the approximation and generalization rates are governed by the interaction between the input dimensions, regularities, and Sobolev orders, while the dependence on the model complexity retains a \(\log\log/\log\)-type structure. Our analysis provides a general theoretical framework for multi-input operator learning, including Sobolev training, and is applicable to operator learning problems arising from partial differential equations and scientific computing.

07.
arXiv (CS.AI) 2026-06-19

Calibration Without Comprehension: Diagnosing the Limits of Fine-Tuning LLMs for Vulnerability Detection in Systems Software

arXiv:2606.20502v1 Announce Type: cross Abstract: Whether LLMs scoring well on vulnerability benchmarks genuinely reason about security or merely pattern-match on contaminated data remains unresolved. We present CWE-Trace, a framework for LLM vulnerability detection built from 834 manually curated Linux kernel samples spanning 74 CWEs. The framework enforces a strict temporal split (pre-2025 historical set / post-cutoff leakage-free set), preserves context-aware vulnerable–patched pairs, and introduces two diagnostic metrics: the Directional Failure Index (DFI) and Hierarchical Distance and Direction (HDD). We evaluate eight vanilla LLMs and 15 LoRA fine-tuned variants across non-targeted detection, targeted detection, and CWE classification. Our analysis yields two key results. First, data contamination provides no measurable advantage. Function-level analysis shows that 84% of nominally contaminated samples carry no usable memorization signal: vulnerable functions are absent or cross-mapped across datasets, and ~31% of contaminated samples carry CWE misclassification. Second, backbone directional priors dominate fine-tuning. Models exhibit stable, systematic failure modes (DFI ranging from -85.5 to +94.8 pp) that persist from historical to post-cutoff data and resist correction. Fine-tuning shifts the output threshold without changing the decision policy. This is calibration without comprehension: output distributions adapt to training data while the underlying security reasoning remains absent. The weakest backbone at binary detection (DeepSeek-R1) gains the most in coarse CWE classification, revealing that detection and understanding are decoupled capabilities. The best detection score reaches only 52.1% (+2.1 pp above chance); exact CWE ranking remains below 1.3% Top-1 accuracy, confirming that current LLMs lack reliable security reasoning for systems software, regardless of fine-tuning strategy.

08.
bioRxiv (Bioinfo) 2026-06-18

MorphoStat: A Statistics-Aware Pipeline for Morphological Profiling Analysis

Authors:

High-content imaging produces thousands of morphological measurements per cell. Interpreting these measurements requires normalization to remove plate effects, statistical tests selected on the basis of data distribution, and control over false discoveries across many features tested at once. MorphoStat is an open-source Python pipeline that applies this sequence of steps automatically. Given a CSV file from CellProfiler or a compatible imaging platform, it removes low-quality wells, normalizes each plate against DMSO controls using a MAD-scaled z-score, routes each feature to a parametric or nonparametric test based on a distributional check, applies Benjamini Hochberg correction, and writes out results and publication-ready figures. On the BBBC021 benchmark (MCF-7 breast-cancer cells, 632 wells, 473 features), MorphoStat recovered 12 of 13 known mechanism-of-action classes in principal component space, confirming that the normalization and statistical routing work as intended. The tool is available at https://github.com/Almunthir334/morphostat (DOI: 10.5281/zenodo.20354069) under the MIT license.

09.
medRxiv (Medicine) 2026-06-18

Expert in Ultrasound Skills: Feasibility of an IMU-video platform to describe technical profiles during focused cardiac ultrasound. Pilot study

Background: Focused cardiac ultrasound (FoCUS) is operator dependent and requires coordinated probe manipulation, image interpretation and iterative visual feedback. Existing assessment approaches often emphasize final image quality or expert rating. We developed Expert in Ultrasound Skills (EXUS) , a platform that synchronizes transducer-mounted inertial measurement unit (IMU) data with ultrasound video, and evaluated its technical feasibility during FoCUS acquisition. Methods: This observational pilot study included 6 operators performing two repetitions of a four-view FoCUS protocol, yielding 12 analytical sessions and 48 planned acquisitions. Feasibility was defined by acquisition completion, video availability, start/stop events, fused IMU-video windows, temporal coverage, complete human label entries and IMU integrity. A 100-image Likert rating task was used to summarize pairwise inter-rater agreement for still-frame image quality assessment. Results: All 48 planned acquisitions were completed with video, start/stop events, fused windows and complete human label entries. Temporal coverage was at least 90% in 47/48 acquisitions. IMU integrity endpoints exceeded the 80% threshold: 43/48 acquisitions had no extreme IMU-derived artifact, 43/48 had no active-segment IMU restart and 44/48 had no complete motion flatline. Mean pairwise exact agreement for the Likert task was 38.9%, with mean quadratic-weighted Cohen's kappa of 0.564. Post hoc profiles varied across duration, visual quality, mechanical load and motor efficiency. Conclusions: EXUS was technically feasible for synchronized IMU-video capture during FoCUS. The pilot supports multimodal acquisition data as a way to describe technical profiles and generate formative feedback hypotheses, but the post hoc indices are not validated competency measures. Keywords: focused cardiac ultrasound; point-of-care ultrasound; inertial measurement unit; medical education; deliberate practice

10.
arXiv (CS.AI) 2026-06-19

See-and-Reach: Precise Vision-Language Navigation for UAVs within the Field of View

arXiv:2606.20045v1 Announce Type: cross Abstract: UAV Vision-Language Navigation (UAV-VLN) is typically formulated as a holistic search-and-reach problem, where long-range target discovery and final target approach are optimized and evaluated jointly. This formulation makes it difficult to assess a critical capability of aerial embodied agents, namely whether a UAV can accurately ground a visible target and translate vision-language evidence into precise 3D motion once the target enters its field of view. To address this limitation, we introduce UAV-VLN-FOV, a target-visible navigation task that isolates the see-and-reach stage and enables a more diagnostic evaluation of terminal reaching ability. We further propose 3DG-VLN, a vision-language waypoint prediction framework guided by dynamic 3D direction cues to enhance fine-grained visual grounding and spatial direction alignment for precise target reaching. Specifically, 3DG-VLN adaptively processes high-resolution front-view and downward-view observations to preserve fine-grained visual and geometric details for target grounding. It also updates the target-relative direction online during closed-loop navigation, allowing the agent to maintain spatial alignment with the target and reduce accumulated direction drift. To support this task, we construct a dedicated high-resolution benchmark which contains 2,717 trajectories with target-oriented high-level instructions, high-resolution front-view and downward-view egocentric observations, and continuous 3D waypoint annotations. Experiments show that 3DG-VLN outperforms competitive UAV-VLN baselines, achieving a 13.82\% improvement in success rate. Real-world trials further demonstrate the potential of 3DG-VLN for practical see-and-reach navigation. The source code and benchmark are available at https://github.com/xuefanfu/3DG-VLN.

11.
Nature (Science) 2026-06-17

Revealing competitive interfacial reactions in high-energy Li–S batteries

Authors:

Charge transfer at solid–liquid interfaces plays a critical role in various energy-storage systems1, particularly under dynamically varying reactant concentrations. Deciphering these intricate reaction pathways remains a substantial challenge, notably in lithium–sulfur (Li–S) batteries, in which achieving high energy density requires efficient conversion of highly concentrated lithium polysulfides (LiPSs)2,3. However, the mechanisms governing lithium sulfide (Li2S) deposition and dissolution under lean electrolyte conditions remain poorly understood. Here, using in situ liquid-cell electron microscopy, we directly visualize concentration-driven phase segregation at the electrode–electrolyte interface. Within these high-concentration interfacial layers (HCILs), competitive surface and solution dictate the charge-transfer dynamics and ultimately govern Li2S deposition at different phase boundaries. Density functional theory (DFT) calculations reveal that the aggregation of LiPSs alters molecular geometry, electronic properties and orbital hybridization, collectively facilitating charge transfer through highly concentrated LiPSs clusters. Guided by these insights, we design optimized electrodes that balance interfacial reaction pathways, enabling fast charging (4 C, 26.8 mA cm−2) and achieving high energy densities exceeding 400 Wh kg−1. These findings provide mechanistic understanding of interfacial reactions under practical working conditions and offer a design strategy to advance Li–S batteries. Visualization of concentration-driven phase segregation within high-concentration interfacial layers in the context of high-energy lithium–sulfur batteries using liquid-cell electrochemical transmission electron microscopy reveals competitive interfacial reactions under lean electrolyte conditions at different phase boundaries.

12.
arXiv (CS.CL) 2026-06-16

Scaling Human and G2P Supervision for Robust Phonetic Transcription

Expert phonetic annotation is costly, especially for non-standard dialects and atypical speech. A common alternative is using Grapheme-to-Phoneme (G2P) models to auto-generate phonetic labels from text transcripts at scale. We study how automatic phonetic transcription performance scales with human and G2P supervision in English. Using a curated 80-hour benchmark spanning native, non-native and post-stroke speech, we identify a supervision quality threshold: G2P supervision helps only when fewer than 20-30 hours of human annotation are available. Beyond this threshold, it provides no significant benefit and can reduce cross-dialect robustness. What is effective after this threshold is ASR pretraining which we use to achieve a 2.3x reduction in weighted phone feature error rate over prior systems, with strong gains on non-native and aphasic speech. These results suggest that quantity-driven G2P scaling may yield diminishing returns for robust generalization.

13.
arXiv (CS.LG) 2026-06-16

Task-Error Residual Learning for Real-Robot Five-Ball Juggling

arXiv:2606.16978v1 Announce Type: cross Abstract: For residual learning that refines existing behavior, sample efficiency depends on two things: how much information each rollout returns, and how efficiently the learner uses that information. Reinforcement learning's standard scalar reward carries far less information than the directional task error that defines the task. Random exploration further discards whatever information each rollout returns. Through residual learning with directional task-error supervision and a task error model that drives sample selection, we achieve stable three-, four-, and five-ball juggling on anthropomorphic Barrett WAM arms. Despite planning and controlling through a simple, idealized stack, the system converges from the second attempt. The first attempt drops, after which task error decreases monotonically without further failures. In comparison, five-ball juggling typically takes humans years of practice. We compare residual learners across two ternary axes, the directional information in the learning feedback and the commitment of the analytic prior, spanning Newton-style Jacobian updates, Composite Bayesian Optimization, and stochastic search methods. Both axes prove necessary: neither directional feedback nor an informative prior suffices alone, and the simplest method that combines them, a fixed-Jacobian Newton update, is the most reliable. The learned residual tolerates substantial prior misalignment and degraded joint tracking, affecting mainly convergence speed. The bottleneck for residual learning on real robots is therefore the information content of the supervision signal and how the learner uses it, not the accuracy of the surrounding stack. Video documentation of all experiments is available at https://kai-ploeger.com/residual-juggling.

14.
arXiv (CS.LG) 2026-06-16

CREST: Deployment-Realistic Hardware-in-the-Loop NAS for Embedded Sensing Systems

arXiv:2606.15004v1 Announce Type: cross Abstract: Deploying neural networks on low-power microcontrollers (MCUs) requires selecting model architectures under tight memory, latency, and energy constraints. Existing workflows often simplify this process along one or more axes: static proxy costs such as FLOPs or parameters, treating one MCU as representative, and continuous-inference tests instead of deployed sensing schedules. These assumptions can mis-rank Pareto-front candidates, miss infeasible deployments, and obscure schedule-dependent energy. We present CREST (Cross-platform Runtime Evaluation and Search Tool), a deployment-realistic hardware-in-the-loop (HIL) neural architecture search (NAS) framework for MCU sensing systems. CREST keeps the optimizer, HIL measurement boundary, logging, and replay workflow fixed while exposing workload, model family, target backend, schedule, quantization, and scoring policy as configurable axes. This makes deployment effects experimentally separable within one reusable workflow. We evaluate CREST on inertial odometry and audio classification across three Arm Cortex-M targets. For inertial odometry, measured-energy HIL search reduces median per-inference energy by 41.7% versus FLOPs-based selection and 40.8% versus memory-traffic-based selection at similar error. FLOPs-based selection also chooses infeasible deployments on memory-constrained targets. On the STM32 N657 target, continuous-inference and duty-cycled searches produce different Pareto frontiers. For audio classification, the same application-level policy selects different DS-CNN architectures on different boards, and cross-board replay changes deployment cost substantially. Overall, CREST shows that deployment-realistic MCU NAS must jointly optimize model architecture, target platform, runtime schedule, and deployment policy rather than relying only on static proxy costs or continuous-inference measurements.

15.
arXiv (CS.AI) 2026-06-19

The Tao of Agency: Autotelic AI, Embedded Agency and Dissolution of the Self

Authors:

arXiv:2606.19924v1 Announce Type: new Abstract: Most artificial intelligence systems are built on the assumption that goals are exogenous and specified by the designer. Exploring what happens when an agent begins generating its own goals opens the field of autotelic AI. Agents are expected not merely to pursue objectives but to discover them. In this article, we trace its consequences through intrinsic motivation, resource-driven priors, causal-interventional learning, homeostasis, and embeddedness; the last of which is found to be a necessary but not sufficient condition for autotelic agency. Embeddedness individuates the agent at the cost of revealing that the individuation is non-unique, such that the same dynamics admit many valid partitions, each defining a different candidate self. The deepest problem with autotelic AI is therefore not how the agent generates goals, but how it generates and relativizes the self to which the goals are assigned. The agent must believe in its own boundary in order to act, and see through that boundary in order to understand. We consolidate these developments into a single framework and extend it along three directions: a quantum formulation in which the agent-environment cut becomes physical, a philosophical reading against non-dual contemplative traditions, and a concrete LLM-based agentic instantiation.

16.
arXiv (CS.AI) 2026-06-19

CADBench: A Multimodal Benchmark for AI-Assisted CAD Program Generation

arXiv:2605.10873v2 Announce Type: replace-cross Abstract: Recovering editable CAD programs from images or 3D observations is central to AI-assisted design, but progress is difficult to measure because existing evaluations are fragmented across datasets, modalities, and metrics. We introduce CADBench, a unified benchmark for multimodal CAD program generation. CADBench contains 18,000 evaluation samples spanning six benchmark families derived from DeepCAD, Fusion 360, ABC, MCB, and Objaverse; five input modalities including clean meshes, noisy meshes, single-view renders, photorealistic renders, and multi-view renders; and six metrics covering geometric fidelity, executability, and program compactness. STEP-based families are stratified by B-rep face count and all families are diversity-sampled to support controlled analysis across complexity and object variation. We benchmark eleven CAD-specialized and general-purpose vision-language systems, generating more than 1.4 million CAD programs. Under idealized inputs, specialized mesh-to-CAD models substantially outperform code-generating VLMs, which remain far from reliable CAD program reconstruction. CADBench further reveals three recurring failure modes: reconstruction quality degrades with geometric complexity, CAD-specialized models can be brittle under modality shift, and model rankings change across metrics. Together, these results position CADBench as a diagnostic testbed for measuring progress in editable 3D reconstruction and multimodal CAD understanding. The benchmark is publicly available at https://github.com/anniedoris/CADBench.

17.
arXiv (CS.LG) 2026-06-17

MGUP: A Momentum-Gradient Alignment Update Policy for Stochastic Optimization

arXiv:2606.17526v1 Announce Type: new Abstract: Efficient optimization is essential for training large language models. Although intra-layer selective updates have been explored, a general mechanism that enables fine-grained control while ensuring convergence guarantees is still lacking. To bridge this gap, we propose MGUP, a novel mechanism for selective updates. MGUP augments standard momentum-based optimizers by applying larger step-sizes to a selected fixed proportion of parameters in each iteration, while applying smaller, non-zero step-sizes to the rest. As a nearly {plug-and-play} module, MGUP seamlessly integrates with optimizers such as AdamW, Lion, and Muon. This yields powerful variants such as MGUP-AdamW, MGUP-Lion, and MGUP-Muon. Under standard assumptions, we provide theoretical convergence guarantees for MGUP-AdamW (without weight decay) in stochastic optimization. Extensive experiments across diverse tasks, including MAE pretraining, LLM pretraining, and downstream fine-tuning, demonstrate that our MGUP-enhanced optimizers achieve superior or more stable performance compared to their original base optimizers. We offer a principled, versatile, and theoretically grounded strategy for efficient intra-layer selective updates, accelerating and stabilizing the training of large-scale models. The code is publicly available at https://github.com/MaeChd/MGUP.

18.
arXiv (CS.AI) 2026-06-16

UrbanWell: Benchmarking Multimodal Large Language Models for Spatio-Temporal Urban Wellbeing Analytics

arXiv:2606.15890v1 Announce Type: new Abstract: Understanding urban wellbeing from multimodal data requires integrating heterogeneous spatial and temporal signals, posing significant challenges for current multimodal large language models (MLLMs). We introduce UrbanWell, a large-scale benchmark designed to systematically evaluate the spatio-temporal reasoning capabilities of MLLMs for urban wellbeing analytics through joint modeling of satellite and street view imagery. UrbanWell spans 38 cities across multiple years and includes diverse indicators covering (1) environmental conditions (CO$_2$, NO$_2$, PM${2.5}$, and Normalized Difference Vegetation Index), (2) spatial accessibility (minimum distance to supermarkets and restaurants), (3) urban form (road length, road density, and land use), (4) urban vitality (population, economic activity diversity, and land use diversity), and (5) subjective perception attributes (e.g., safety, beauty, liveliness, wealth, and quietness). All indicators are aligned at grid level to enable standardized evaluation. Beyond static prediction, UrbanWell defines temporal reasoning tasks, including future value forecasting from historical observations and temporal trend classification. We benchmark 15 state-of-the-art representative MLLMs in a zero-shot setting, providing a comprehensive comparative evaluation across spatial and temporal dimensions. Experimental results indicate that while MLLMs capture salient spatial and perceptual cues, their performance varies substantially across heterogeneous urban indicators spanning environment and subjective perception. UrbanWell serves as a unified benchmark for evaluating multimodal spatial and temporal reasoning in urban wellbeing analytics, offering a standardized testbed for systematic assessment and future research on multimodal urban intelligence. Our codes and datasets are accessible via https://github.com/axin1301/UrbanWell-Benchmark.

19.
arXiv (CS.CL) 2026-06-18

From Concept-Aligned Tokens to Vulnerable Features: Mechanistic Localization of Jailbreaks

Jailbreak attacks expose a persistent failure mode in safety-aligned LLMs: models can be pushed into harmful behavior, but the internal representations enabling this shift remain poorly localized. Recent mechanistic safety studies often explain such behavior through broad representational objects, including global refusal directions, activation steering vectors, and refusal-related SAE features. We instead ask whether jailbreak vulnerability can be traced to finer-grained, prompt-conditioned SAE feature subgroups. We introduce a token-driven mechanistic pipeline that decomposes the residual stream of Gemma-2-2B into Sparse Autoencoder (SAE) features and identifies feature subgroups associated with unsafe behavior. Using single-category unsafe examples from BeaverTails to reduce cross-category interference, we extract harmful concepts from adversarial responses and align them with concept-relevant prompt tokens through subspace similarity. We then apply three feature-grouping strategies: cluster-based, hierarchical-linkage, and single-token-driven, to identify SAE feature subgroups across all 26 layers. Finally, we amplify the top features in each subgroup and evaluate the resulting generations with a standardized harmfulness judge. Single-token-driven grouping achieves harmfulness comparable to full cluster-based grouping, showing that individual harmful prompt tokens are sufficient to localize vulnerability-relevant SAE feature subgroups without relying on broader cluster-level aggregation. These subgroups appear across early and mid-to-late layers, with stronger concentration in mid-to-late layers, where targeted steering exposes specific model vulnerabilities. Overall, our results suggest that jailbreak susceptibility can be traced to sparse, token-localized SAE feature subgroups, complementing prior accounts based on broad adversarial, refusal, or steering directions.

20.
arXiv (CS.LG) 2026-06-19

The Representational Limit of Scalar Interactions: An Interventional Decomposition

arXiv:2606.19410v1 Announce Type: cross Abstract: Signed pairwise interaction scores fundamentally conflate uniqueness (U), redundancy (R), and synergy (S). We prove this on a minimal 3-way XOR structural causal model: faithful indices such as Shapley-Taylor return zero per pair, whereas projective indices such as Shapley Interaction spread the third-order effect into pair scalars that conflate the three mechanisms. We introduce Stochastic Hi-Fi, a post-hoc, retraining-free predictability decomposition that estimates per-feature U/R/S profiles by interventional masked inference. The estimator provides exact interventional semantics, finite-sample Monte Carlo bounds, strict variance reduction from coupled diamond sampling, and uniform finite-vocabulary convergence. Across tabular SCMs, Stochastic Hi-Fi recovers structure missed by scalar baselines (up to 411x larger interaction-magnitude recovery ratios). It also separates redundant and synergistic heads in the GPT-2 IOI circuit. On NIH ChestX-ray14, Stochastic Hi-Fi matches GradCAM on Pointing Game and improves substantially on Deletion AUC.

21.
arXiv (CS.LG) 2026-06-11

Deep Learning of Solver-Aware Turbulence Closures from Nudged LES Dynamics

arXiv:2604.23874v3 Announce Type: replace-cross Abstract: The differentiable physics paradigm may be leveraged as an a-posteriori approach for discovering turbulence closure models by embedding a neural network parameterization directly inside the solver and optimizing it given potentially sparse target data. This addresses a key limitation of a-priori learning where direct numerical simulation (DNS) data is used to approximate the subgrid stress with the assumption of a low-pass filter. Closures trained in this a-priori manner frequently lead to unstable deployments due to the mismatch between the assumed filter and the effect of numerical discretizations and coarse-graining. In comparison, while typically stable during deployment, a-posteriori learning incurs high computational costs due to the need to backpropagate through a large eddy simulation (LES) solver. Furthermore, a-posteriori methods are challenging to apply broadly since they require significant modification of existing solvers. Finally, both approaches are limited when generalization is desired across different numerical schemes with their implicit filtering characteristics. In this work, we present a deep-learning approach for turbulence closure modeling built on the continuous data assimilation framework. Our approach enables the a-priori training of closures using sparsely observed DNS data without modifying or differentiating through the LES solver, while preserving stability during deployment for the recovery of invariant statistics. We focus on the model's ability to adapt to different discretizations by explicitly conditioning it on the numerical scheme. We use two- and three-dimensional canonical cases to test our framework and show that the learned correction systematically tracks the discretization error of the coarse solver.

22.
arXiv (CS.AI) 2026-06-16

AgentLeak: A Benchmark for Internal-Channel Privacy Leakage in Multi-Agent LLM Systems

arXiv:2602.11510v3 Announce Type: replace Abstract: Multi-agent Large Language Model (LLM) systems create privacy risks that current output-only benchmarks cannot measure. When agents coordinate on tasks, sensitive data may pass through inter-agent messages, shared memory, and tool arguments, all pathways that final-output audits typically do not inspect. We introduce AgentLeak, a benchmark for evaluating internal-channel privacy leakage in multi-agent LLM systems. AgentLeak instruments seven privacy-relevant communication pathways and provides a large-scale empirical evaluation focused on final outputs, inter-agent messages, and shared memory. Across 1,000 scenarios spanning healthcare, finance, legal, and corporate domains, five production LLMs (GPT-4o, GPT-4o-mini, Claude 3.5 Sonnet, Mistral Large, and Llama 3.3 70B), and 4,979 validated execution traces, we find that multi-agent configurations reduce final-output leakage (C1: 27.2% vs 43.2% in single-agent mode) compared with single-agent baselines but introduce internal channels that raise total system exposure to 68.9% (aggregated across C1, C2, C5). Inter-agent messages (C2) leak at 68.8%, compared with 27.2% for final outputs (C1), meaning that output-only audits miss 41.7% of violations. Across all five models and four domains, the pattern C2 $\geq$ C1 holds consistently. These results suggest, within the evaluated coordinator-worker setting, that privacy risk in multi-agent systems is strongly shaped by architectural coordination channels rather than final-output behavior alone: it arises from internal channels that remain invisible to standard output-level defenses.

23.
arXiv (CS.CL) 2026-06-15

Deja Vu at Scale: Paraphrase-Robust Detection of Duplicate Gherkin Steps in Behaviour-Driven Software Testing with Sentence-Transformer Embeddings and a 1.1M-Step Open Benchmark

Context. Behaviour-Driven Development (BDD) suites in Gherkin accumulate step-text duplication with documented maintenance cost. Prior detectors either require runnable tests or are single-organisation, leaving a gap: a static, paraphrase-robust, step-level detector and a public benchmark to calibrate it. Objective. We release (i) the largest cross-organisational BDD step corpus to date, (ii) a labelled pair-level calibration benchmark, and (iii) a four-strategy detector with a consolidation-savings model linking clusters to ISO/IEC 25010 maintainability sub-characteristics. Method. The corpus contains 347 public GitHub repositories, 23,667 .feature files, and 1,113,616 Gherkin steps, SPDX-tagged. The detector layers exact hashing, normalised Levenshtein, sentence-transformer cosine, and a Levenshtein-banded hybrid. Calibration uses 1,020 manually labelled step pairs under a released rubric (60-pair overlap, Fleiss kappa = 0.84). We report precision, recall, and F1 with bootstrap 95% CIs under the primary rubric and a score-free relabelling, and benchmark against SourcererCC-style and NiCad-style lexical baselines. Results. Step-weighted exact-duplicate rate is 80.2%; median-repository rate is 58.6% (Spearman rho = 0.51). The top hybrid cluster has 20,737 occurrences across 2,245 files. Near-exact reaches F1 = 0.822 on score-free labels; semantic F1 = 0.906 under the primary rubric reflects a disclosed stratification artefact. Lexical baselines reach F1 = 0.761 and 0.799. The savings model estimates 893,357 corpus-wide eliminable step occurrences; on the median repository 62.5% of step lines are eliminable.

24.
arXiv (CS.LG) 2026-06-12

Generative Modeling of Bach-Style Symbolic Music: A Comparative Study of Autoregressive, Latent-Variable, and Adversarial Approaches

arXiv:2606.13626v1 Announce Type: cross Abstract: We study generative modeling of Bach-style symbolic piano music using a shared MIDI corpus and three model families: autoregressive LSTMs with attention, latent-variable models including recurrent VAEs and vector-quantized VAEs, and generative adversarial networks. We compare their ability to model polyphonic note sequences, learn useful latent representations, and generate stylistically coherent compositions. Our experiments show that the autoregressive LSTM with attention produces the most musically coherent samples, while vector quantization helps mitigate posterior collapse and yields more structured outputs than conventional recurrent VAEs. The adversarial approach captures local pitch patterns but remains difficult to train and generalizes less reliably to Bach's style. These results highlight the relative strengths and failure modes of autoregressive, latent-variable, and adversarial approaches for symbolic music generation.

25.
arXiv (math.PR) 2026-06-15

Mixing Times for the Facilitated Exclusion Process

arXiv:2402.18999v2 Announce Type: replace Abstract: The facilitated simple exclusion process (FEP) is a one-dimensional exclusion process with a dynamical constraint. We establish bounds on the mixing time of the FEP on the segment, with closed boundaries, and the circle. The FEP on these spaces exhibits transient states that, if the macroscopic density of particles is at least $1/2$, the process will eventually exit to reach an ergodic component. If the macroscopic density is less than $1/2$ the process will hit an absorbing state. We show that the symmetric FEP (SFEP) on the segment $\{1,\ldots,N\}$, with $k>N/2$ particles, has mixing time of order $N^{2}\log(N-k)$ and exhibits the pre-cutoff phenomenon. For the asymmetric FEP (AFEP) on the segment, we show that there exists initial conditions for which the hitting time of the ergodic component is exponentially slow in the number of holes $N-k$. In particular, when $N-k$ is large enough, the hitting time of the ergodic component determines the mixing time. For the SFEP on the circle of size $N$, and macroscopic particle density $\rho \in(1/2,1)$, we establish bounds on the mixing time of order $N^{2}\log N$ for the process restricted to its ergodic component. We also give an upper bound on the hitting time of the ergodic component of order $N^{2}\log N$ for a large class of initial conditions. The proofs rely on couplings with exclusion processes (both open and closed boundaries) via a novel lattice path (height function) construction of the FEP.