Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
medRxiv (Medicine) 2026-06-15

ICD-10 Code Ambiguity Obscures Treatment-Eligible Adults with Spinal Muscular Atrophy: A Single-Center Chart Review and Patient Outreach Study

Background. Three disease-modifying therapies (DMTs) for spinal muscular atrophy (SMA) have been approved since 2016, yet many adults remain untreated. Identifying them depends on ICD-10 codes that capture SMA but do not reliably distinguish it from other related conditions. We examined, in one U.S. health system, both patients' engagement with therapy and the accuracy of the codes used to find them. Methods. We conducted a retrospective chart review of adults in an academic health system identified by SMA-associated ICD-10 codes, with manual adjudication of diagnosis and DMT status. Confirmed SMA-positive, DMT-naive patients were invited to a structured telephone interview on treatment awareness and barriers. Results. Of 60 charts, 22 (36.7%; 95% CI 25.6-49.3%) were appropriately coded for SMA or a related disorder; only 16 (26.7%) had molecularly confirmed SMA. The other 38 (63.3%) were miscoded, spanning spinal and bulbar muscular atrophy, asymptomatic carriers, prenatal screening, and conditions unrelated to SMA. Ten of the 16 confirmed patients (62.5%) were DMT-naive; one was interviewed, one declined, and eight could not be reached. The non-response is itself a finding: the patients least visible to administrative data are the hardest to reach. Conclusions. ICD-10 ambiguity is a barrier to treatment access in adult SMA, as is loss to follow-up. We make two recommendations: continuous documentation-coding alignment that uses natural language processing to verify the genetic precondition, and type-specific SMA codes (subcodes for Types 0-4) anchored on molecular SMN1 confirmation. Together these would support cohort identification, outreach, and evidence generation without adding to clinician burden.

02.
arXiv (CS.AI) 2026-06-16

Bridging the Gap: Enabling Natural Language Queries for NoSQL Databases through Text-to-NoSQL Translation

arXiv:2502.11201v3 Announce Type: replace-cross Abstract: NoSQL databases are core data infrastructure, yet natural-language access to them remains underdeveloped: correct query generation must recover how a non-relational data model represents entities, nested paths, arrays, missing fields, and dynamic keys. This paper studies Text-to-NoSQL, translating natural-language requests into executable NoSQL queries, instantiated with MongoDB aggregation pipelines over schema-less document stores. We present TEND, short for Text-to-NoSQL Dataset, an execution-verified benchmark with 1,210 MongoDB-native tasks across 11 databases. To our knowledge, TEND is the first Text-to-NoSQL benchmark whose database worlds are MongoDB-native by design: experts manually define collection boundaries, nested arrays, optional and sparse paths, polymorphic shapes, and dynamic-key conventions; these worlds are populated with real data and verified through frozen MongoDB execution, so TEND evaluates schema-less document reasoning rather than SQL-to-MQL transfer. We further introduce SAG, a Schema-as-Data Grounding solver that induces path and value grounding from stored-document evidence before bounded MQL generation, execution-grounded repair, and result-consistency selection. Evaluation uses bounded column-tolerant execution accuracy (EXC) as the headline metric, complemented by a graded result-set F1 and a mutually exclusive execution-outcome decomposition. Experiments show that LLMs with strong NL2SQL performance degrade substantially on TEND, validating Text-to-NoSQL as a distinct schema-less document reasoning problem.

03.
arXiv (CS.AI) 2026-06-15

Hy-Embodied-0.5-VLA: From Vision-Language-Action Models to a Real-World Robot Learning Stack

arXiv:2606.14409v1 Announce Type: cross Abstract: In this report, we present Hy-Embodied-0.5-VLA, abbreviated as HyVLA-0.5, an end-to-end system that spans the full robot learning stack: data collection, model design, continued pre-training and supervised fine-tuning, RL post-training, and real-world deployment. Each component serves a distinct role in this stack.

04.
bioRxiv (Bioinfo) 2026-06-18

Benchmarking attention-based methods for vision transformers' interpretability in retinal fundus imaging

Deep learning models based on Vision Transformers (ViTs) have shown strong performance in retinal fundus imaging, but their interpretability remains poorly understood. In particular, attention-based attribution methods are widely used to explain ViT predictions, despite limited evaluation of their faithfulness and biological relevance in medical imaging. Here, we systematically benchmark four attention-based interpretability methods for RETFound, a retinal ViT-based foundation model, that we previously fine-tuned to predict 17 retinal vascular phenotypes from UK Biobank fundus images1. We compare raw attention, attention rollout, gradient-weighted attention rollout, and Chefer's hybrid relevance-based method using both qualitative visualisation and quantitative evaluation frameworks. To assess attribution faithfulness, we perform perturbation-based deletion and insertion experiments, quantifying changes in model predictions as highly attended image regions are progressively removed or restored. To evaluate biological specificity, we run structure-aware analyses combining attribution maps with vessel segmentation and artery-vein labels through the Relative ratio of Attention Intensity (RAI) metric. Across models, attribution maps differed substantially depending on the selected interpretability method, highlighting the need for rigorous quantitative evaluation. Among the evaluated approaches, gradient-weighted attention rollout consistently achieved the strongest perturbation performance and produced attribution maps most closely aligned with the anatomical definition of the predicted retinal traits. Furthermore, vessel-type specific models systematically concentrate attention on the corresponding vascular structures despite being trained using only a single scalar value per image as supervision. These findings demonstrate that attention-based attribution methods capture biologically meaningful vascular representations, while also revealing method-dependent variability in attribution behaviour. This work provides a quantitative framework for evaluating interpretability methods in medical imaging with annotated segmentation and contributes toward more transparent and biologically grounded medical AI systems.

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

Understanding LLM Reasoning for Abstractive Summarization

Reasoning has substantially improved Large Language Models (LLMs) on analytical tasks such as mathematics and code generation, but its value for abstractive summarization remains unclear. To address this gap, we adapt general reasoning strategies to the summarization setting and conduct a large-scale comparative study of 8 reasoning strategies and 3 Large Reasoning Models (LRMs) across 8 diverse datasets, evaluating both summary quality and factual faithfulness. Our results show that reasoning is not a universal solution and its effectiveness depends strongly on the strategy and the summarization setting. In particular, we find a trade-off between summary quality and factual faithfulness. Explicit reasoning strategies often improve reference-based quality, but may weaken factual grounding, whereas implicit reasoning in LRMs shows the opposite tendency. We further find that increasing an LRM's internal reasoning budget does not reliably improve summarization and can even reduce factual consistency. These findings suggest that, for summarization, more reasoning is not always better. Effective reasoning should preserve faithful compression rather than induce over-elaboration. Our source code is publicly available.

07.
arXiv (CS.AI) 2026-06-12

Lightweight and Interpretable Transformer via Mixed Graph Algorithm Unrolling for Traffic Forecast

arXiv:2505.13102v4 Announce Type: replace-cross Abstract: Unlike conventional "black-box" transformers with classical self-attention mechanism, we build a lightweight and interpretable transformer-like neural net by unrolling a mixed-graph-based optimization algorithm to forecast traffic with spatial and temporal dimensions. We construct two graphs: an undirected graph $\mathcal{G}^u$ capturing spatial correlations across geography, and a directed graph $\mathcal{G}^d$ capturing sequential relationships over time. We predict future samples of signal $\mathbf{x}$, assuming it is "smooth" with respect to both $\mathcal{G}^u$ and $\mathcal{G}^d$, where we design new $\ell_2$ and $\ell_1$-norm variational terms to quantify and promote signal smoothness (low-frequency reconstruction) on a directed graph. We design an iterative algorithm based on alternating direction method of multipliers (ADMM), and unroll it into a feed-forward network for data-driven parameter learning. We periodically insert graph learning modules for $\mathcal{G}^u$ and $\mathcal{G}^d$ that play the role of self-attention. Experiments show that our unrolled networks achieve competitive traffic forecast performance as state-of-the-art prediction schemes, while reducing parameter counts drastically.

08.
arXiv (CS.LG) 2026-06-19

Semantic-Anchored Evidential Fusion for Domain-Robust Whole-Slide Survival Analysis

arXiv:2606.19966v1 Announce Type: cross Abstract: Whole-slide images (WSIs) are widely used for computational cancer prognosis. However, most existing methods primarily focus on in-domain performance and fail to generalize across clinical centers. This limitation stems from their reliance on pixel-derived representations that are highly susceptible to domain-specific artifacts caused by staining protocols and scanner hardware. We hypothesize that high-level pathology semantics, such as tumor grade and micro-environmental architecture, provide a domain-invariant semantic representation that mirrors the robust diagnostic logic of human pathologists. Therefore, we propose a Semantic-Anchored Evidential Fusion Survival (SAEFS) framework, where SAEFS derives semantic anchors from WSIs via Visual Question Answering (VQA), employs a dual-stream WSI evidence extraction architecture, uses Dirichlet-based Subjective Logic to model uncertainty, and fuses semantic and visual evidence through a cautious conjunction rule to avoid overconfident fusion from correlated sources. Trained exclusively on one source domain and evaluated zero-shot across four unseen domains, SAEFS consistently outperforms state-of-the-art models both in prediction accuracy and reliability, improving the average C-index by 10.2%. Quantitative analyses further show that VQA-derived semantic features exhibit significantly lower cross-center divergence than pixel-derived features, highlighting their robustness for cross-center clinical applications.

09.
arXiv (quant-ph) 2026-06-25

Tensor network characterization and mitigation of readout errors

arXiv:2606.25974v1 Announce Type: new Abstract: Readout errors are a major bottleneck to extracting reliable information from near-term quantum processors, especially when spatial correlations are non-negligible. We present a unified tensor-network framework that models the readout process as a matrix product operator (MPO), enabling efficient characterization and mitigation beyond uncorrelated approximations. The MPO model is trained via likelihood optimization on calibration data and applies to multiple tasks, including nonlocal observable estimation, random circuit sampling, and random-measurement protocols, such as classical shadows and learning-based tomography. Experiments on a superconducting processor and numerical simulations up to 20 qubits show that the MPO model captures correlated readout errors that uncorrelated models miss, with a sample cost that grows only near-linearly with system size. When extended to two-dimensional systems, the framework can also be integrated with tensor-network quantum error-correction decoders by performing joint inference over data and readout errors. These results establish tensor-network readout error mitigation as a scalable and versatile approach for noise-aware quantum data processing.

10.
arXiv (CS.CV) 2026-06-11

Ouroboros-Spatial: Closing the Data-Model Loop for Spatial Reasoning

Spatial reasoning remains a persistent challenge for multimodal large language models (MLLMs). Existing approaches largely rely on large-scale, statically curated datasets, where all training samples are treated uniformly regardless of the model's evolving capabilities. This static paradigm is inherently data-inefficient: training capacity is often spent on samples that are either trivial or overly difficult for the model at its current stage. To address this limitation, we propose Ouroboros-Spatial, a self-evolving training framework in which the model plays dual roles as a proposer and a solver. In each iteration, a frozen proposer generates spatial question-answer (QA) pairs from 3D scene metadata and raw video frames, together with executable code for deriving reliable ground truth. A learnable solver is then fine-tuned on the accepted samples, and its per-sample prediction confidence is used as a difficulty signal. This signal is fed back to the proposer in the next iteration, guiding it to generate questions better matched to the solver's current capabilities. Through this closed-loop design, the training distribution co-evolves with model ability, reducing redundant trivial examples while filtering out ambiguous or uninformative samples with limited learning value. Across six spatial reasoning benchmarks, Ouroboros-Spatial substantially improves Qwen3-VL-4B and Qwen3-VL-8B while using an order of magnitude fewer training examples than recent large-scale curated datasets. On VSI-Bench, it yields absolute gains of 9.9 and 6.8 points for the 4B and 8B models, respectively, enabling both to outperform a wide range of strong open-source and proprietary baselines.

11.
arXiv (CS.CV) 2026-06-24

WorldOlympiad: Can Your World Model Survive a Triathlon?

We introduce WorldOlympiad, a benchmark for diagnosing video-based world models across physical faithfulness, geometric consistency, and interaction fidelity. While existing benchmarks often focus on visual quality, semantic alignment, or short-term temporal coherence, they provide limited insight into whether generated videos obey physical rules, preserve coherent 3D structure, and sustain controllable interactions over long horizons. To address this gap, WorldOlympiad decomposes world-model evaluation into three complementary dimensions. The physical track uses object segmentation and MLLM-as-judge to assess whether generated videos follow interpretable rules in mechanics, thermal phenomena, and material properties. The geometry track reconstructs generated videos with Gaussian splatting and evaluates structural consistency, cross-view coherence, and camera-trajectory alignment. The interaction track assesses whether generated rollouts follow complex action prompts and maintain smooth, coherent transitions across consecutive video chunks. WorldOlympiad further covers three major downstream scenarios, including gaming, robotics, and general real-world videos, capturing diverse challenges from interactive control and embodied manipulation to open-domain motion and camera dynamics. Together, these tracks and scenarios form a scalable and interpretable evaluation suite that exposes failure modes beyond generic video quality. Experiments on state-of-the-art models reveal substantial gaps in physical reasoning, 3D consistency, and long-horizon interaction, underscoring the need for more structured evaluation protocols for generative world models.

12.
PLOS Computational Biology 2026-06-18

Mechanisms underlying spontaneous and evoked calcium responses in oligodendrocyte precursor cells: A modeling investigation

作者:

by Martin Lardy, Leqi Wang, Claire Guerrier, Veronica T. Cheli, Pablo M. Paez, Anmar Khadra Calcium (Ca2+) signaling has emerged as a central regulator of activity-dependent myelination in oligodendrocytes. These Ca2+ signals encompass both the stimulus-independent spontaneous Ca2+ local transients (SCaLTs) generated intrinsically in a voltage-independent manner or facilitated by the membrane voltage, as well as evoked responses triggered by ATP and glutamate release. To investigate the regulatory mechanisms underlying this combined spiking activity, we developed a stochastic spatiotemporal flux-balance model of Ca2+ transients in oligodendrocyte precursor cells (OPCs). The model incorporates all the relevant fluxes in these cells and integrates membrane voltage dynamics with a Ca2+-induced Ca2+-release (CICR) mechanism using parameters fitted to Ca2+ fluorescence recordings. The model reproduced the intrinsic and voltage-facilitated SCaLTs in OPCs in the absence of purinergic and glutamatergic receptors, and captured the three distinct patterns of evoked Ca2+ responses induced by prolonged ATP and glutamate stimulations identified using machine classifier. The model highlighted the role of ATP and glutamate in generating these clusters, and showed that the fast dynamics of CICR is key to producing these evoked responses. Further analysis of the model also revealed that voltage-gated L- and T-type Ca2+ channels slightly increase the frequency of SCaLTs, while stimulation with ATP and glutamate, using randomly distributed pulses mimicking in vivo conditions, leads to an increase in both the amplitudes of Ca2+ spikes (i.e., the combination of SCaLTs and evoked responses) and the prevalence of wide spikes, especially upon glutamate stimulation. Bifurcation analysis of the deterministic version of the model, in the absence of diffusion, demonstrated that ATP and glutamate stimulation can shift the system into an oscillatory regime, thereby increasing the deterministic component of SCaLT dynamics. This study thus offers a comprehensive representation of OPC Ca2+ transients linking recorded in vitro behaviors to in vivo dynamics.

13.
arXiv (CS.CV) 2026-06-12

Quality-Preserving Imperceptible Adversarial Attack on Skeleton-based Human Action Recognition

Adversarial attacks on skeletal human action recognition have received significant attention. However, existing methods typically introduce noise-like perturbations that degrade motion quality post-attack, and thereby are inherently perceptible with recent advancements in S-HAR systems. We discover that this degradation stems from the gap between empirical and true risks during the optimization process of previous adversarial attacks. To address this issue, we propose an attack where adversarial motions are obtained without compromising their motion quality. To minimize the risk gap and preserve motion quality, we propose a distribution-based adversarial attack method without introducing noise-like perturbations. To faithfully evaluate the motion quality, we propose a new metric that aligns with human perception on real-world naturalness. Experiments have been conducted on the state-of-the-art S-HAR methods across two datasets, demonstrating the superiority of our method in both the attack success rate and the post-attack motion quality through qualitative and quantitative analyses. The success of our quality-preserving attack application and distribution-based method raises serious concerns about the robustness of action recognizers, highlighting the need for further enhancements in this domain.

14.
arXiv (CS.CL) 2026-06-15

AdaSR: Adaptive Streaming Reasoning with Hierarchical Relative Policy Optimization

Large reasoning models typically follow a read-then-think paradigm: they observe the complete input, reason over a static context, and then produce the answer. Yet many real-world scenarios are inherently dynamic, such as audio and video stream, where information arrives as a continuous stream and models must reason, update, and respond under partial observations. Recent streaming reasoning methods allow models to think while reading, but they largely rely on supervised imitation of pre-constructed trajectories, which limits their flexibility. In this paper, we propose AdaSR, an adaptive streaming reasoning framework that enables models to reason during input streaming and perform final deliberation once the stream is complete, learning when to think, and how much computation to allocate across different stages. To optimize this hierarchical reasoning process, we introduce Hierarchical Relative Policy Optimization (HRPO), which decomposes policy optimization into streaming reasoning and deep reasoning phases, providing more fine-grained advantage assignment instead of uniformly distributing a single sequence-level advantage over all tokens. HRPO integrates format, accuracy, and adaptive thinking rewards to enforce valid reasoning protocols, preserve final task performance, and encourage latency-aware computation allocation. Experiments show that AdaSR achieves a better balance among reasoning accuracy, computational efficiency, and streaming latency compared with supervised fine-tuning baseline. We release our code at https://github.com/EIT-NLP/StreamingLLM/tree/main/AdaSR.

15.
arXiv (CS.CL) 2026-06-25

Invisible to humans, visible to machines: a preregistered audit of Unicode fidelity across four biomedical bibliographic APIs

Biomedical text mining, scientometrics, and the construction of training corpora for biomedical large language models (LLMs) all assume that the abstract text returned by a bibliographic API faithfully reproduces the published abstract. This pre-registered audit (OSF osf.io/269b5) tests that assumption for four widely used public APIs (PubMed E-utilities, Crossref, OpenAlex, Semantic Scholar) against PubMed Central (PMC) JATS XML as a common ground truth. From a complete enumeration of the PMC Open Access subset for 2024 (about 700,000 records), a simple random sample of 4,000 English-language research articles was drawn; for each, we recorded whether Unicode characters from four pre-specified classes present in the JATS abstract (typographic punctuation, mathematical/scientific symbols, Greek letters, special whitespace) were preserved by each API. Two systematic, deterministic losses met the pre-registered criterion (upper 95% CI bound below 5%): the PubMed AbstractText field preserved typographic punctuation in only 0.6% of eligible abstracts (95% CI 0.3-1.0%), and OpenAlex preserved special whitespace in 0% (0.0-0.4%). A blinded mechanism audit attributed the first loss to character substitution and the second to inverted-index serialization. Mathematical symbols and Greek letters were preserved faithfully (over 95%) by all four APIs. Separately, Crossref returned no abstract for 24.6% of papers (coverage 75.4%, 95% CI 74.1-76.7%), concentrated in specific publishers (Elsevier and ACS: 0%). Character-level fidelity is therefore API-dependent and undocumented: the same publisher-deposited JATS text carries different surface signatures depending on the serving API, with direct consequences for tokenization-sensitive bibliometrics, corpus construction, and character-level indicators of LLM-assisted writing.

16.
arXiv (CS.CV) 2026-06-24

Bridging the Manifold Gap: Riemannian Residual Line Search for One-Step Image Editing

One-step diffusion editors are fast because they avoid inversion and iterative optimization, but a single transport update must be aggressive enough to realize the target prompt and conservative enough to preserve the source image–and no fixed update strength satisfies both demands across edit types. We treat this tension as a post-hoc candidate-selection problem on top of energy-field transport rather than as a new editing model. Our proposed method, Riemannian Residual Line Search, first builds a stronger edit by estimating the local time curvature of the prompt-delta field and projecting the corrected direction back onto the update norm of the original first-order energy-field transport estimation. It then forms a small residual path from the source image to this strong edit, retains the original first-order output as one candidate, and picks the final image by maximizing target-prompt CLIP alignment. On a 700-sample PIE-Bench++ evaluation across 10 edit type IDs, our method achieves state-of-the-art (SOTA) performance among current one-step update algorithms.

17.
arXiv (CS.CV) 2026-06-16

A Multi-Center Benchmark for Abdominal Disease Diagnosis and Report Generation from Non-Contrast CT

Multiphasic contrast-enhanced CT (CECT) is widely used for abdominal lesion characterization, yet it carries inherent risks of contrast-induced nephropathy, escalates acquisition burden, and heavily contributes to radiologist workload. To address these challenges, we introduce a novel multi-center benchmark for multi-organ abdominal disease diagnosis and automated radiology report generation, which learns to synthesize contrast-enhanced findings from single-phase non-contrast CT (NCCT). To support this, we curated a large-scale dataset of paired NCCT-CECT studies and their corresponding contrast-enhanced radiology reports from two centers, partitioned into internal sets and an external validation cohort. Under a unified evaluation protocol, we benchmarked five contemporary deep learning architectures encompassing chest-specific, abdomen-specific, and general-purpose multimodal domains. Extensive experiments demonstrate that NCCT retains diagnostic signals, achieving an average multi-organ AUC of 69.1% on the internal cohort and 63.1% on the external cohort, respectively. By releasing this dataset and standardized benchmark publicly, this study aims to catalyze future research into safer, resource-efficient, and globally accessible contrast-free abdominal imaging workflows. Code is available at: https://github.com/xmed-lab/TriALS-Report.

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

Spotlight: Synergizing Seed Exploration and Spot GPUs for DiT RL Post-Training

arXiv:2606.19004v1 Announce Type: cross Abstract: Reinforcement learning (RL) post-training of Diffusion Transformers (DiTs) is prohibitively expensive, requiring thousands of high-end GPUs. Existing works explore two directions to reduce cost: seed exploration improves training convergence by selecting high-contrast samples, yet adds compute to the critical path; spot GPUs offer 69–77\% lower cost, yet sit idle during training because DiT rollouts finish nearly simultaneously, which prevents LLM-style pipelining of rollout with training. Spot preemptions further break Sequence Parallelism (SP) groups, fragmenting GPU topology. We present Spotlight, the first system that harvests spot GPUs for DiT RL post-training. Spotlight rests on two key insights we devise: (1)~we show that exploration can tolerate stale model weights because exploration that uses the model weights from the previous iteration preserves the relative ranking of random seeds, allowing exploration to run on idle spot GPUs during training. (2)~SP reconfiguration can reuse on-node state, reducing group recovery from minutes to sub-second launches. Built on these insights, Spotlight introduces three techniques: a bandit-based exploration planner that maximizes reward variance within the training time budget, elastic sequence parallelism that reconfigures SP groups on the fly via persistent schedulers and intra-node weight copying, and a preemption-aware pull-based request scheduler that balances load and commits in-flight state upon preemption. We implement Spotlight on the open-source RL platform ROLL and evaluate it on Qwen-Image post-training. Spotlight reaches the same target validation score $4\times$ faster than baselines, reducing total cost by $1.4$-$6.4\times$ while achieving superior image quality on DeepSeek-OCR and Geneval datasets with resolution $512\times512$ and $1280\times1280$.

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

StreamKL: Fast and Memory-Efficient KL Divergence for Boosting Attention Distillation

arXiv:2606.20005v1 Announce Type: cross Abstract: Attention distillation, which trains one attention distribution to match another by minimizing their Kullback-Leibler (KL) divergence, is widely used in knowledge distillation, model compression, continual learning, and sparse-attention LLM training. However, existing approaches materialize both attention distributions before computing the KL reduction, incurring $O(N_QN_K)$ memory and IO costs that become prohibitive at long context lengths. We present StreamKL, the first fused GPU primitive for attention KL divergence that eliminates this quadratic materialization. StreamKL derives a novel online formulation for the coupled two-distribution KL reduction, enabling a single one-pass forward kernel that streams query-key tiles through on-chip SRAM. For the backward pass, StreamKL recomputes attention probabilities tile-by-tile, avoiding storage of quadratic intermediates. We further design and implement efficient GPU kernels with dedicated optimizations. Experiments show StreamKL delivers up to $43\times$ and $14\times$ speedups over baseline methods in the forward and backward passes, respectively. Most importantly, StreamKL reduces the extra HBM footprint of attention distillation from $O(N_QN_K)$ to $O(1)$, enabling long-context distillation on a single GPU.

20.
arXiv (CS.LG) 2026-06-18

Do Time Series Foundation Model Benchmarks Hide Regime-Dependent Failures? Evidence from Traffic Speed Forecasting

arXiv:2606.18367v1 Announce Type: new Abstract: Standard benchmarks evaluate time series foundation models (TSFMs) using aggregate metrics, but these can mask severe failures in critical operating regimes. We introduce regime-stratified evaluation and apply it to three TSFMs on two standard traffic speed benchmarks. Traffic exhibits abrupt regime switching between free-flow and congested states, producing bimodal speed distributions during transitions. When we stratify by traffic regime, both accuracy and prediction-interval coverage degrade sharply during transitions: transition-regime MAE reaches 11 mph (versus 3 mph overall), and empirical coverage of 90% prediction intervals drops as low as 55%. These failures are invisible in aggregate metrics because free-flow observations dominate the sample. A simple historical conditional baseline (sampling from per-sensor training distributions) achieves better transition coverage than any TSFM, but has far worse overall accuracy. We propose bimodal mixture augmentation (BMA), a post-hoc method that combines TSFM forecasts with historical distributional knowledge, approaching the historical baseline's transition coverage while preserving the TSFM's accuracy. Our results suggest that TSFM benchmarks should incorporate regime-aware evaluation to surface failures that aggregate metrics hide.

21.
arXiv (CS.CL) 2026-06-16

Taylor-Calibrate: Principled Initialization for Hybrid Linear Attention Distillation

Hybrid linear attention models offer an appealing path to faster long-context inference: they reduce the quadratic cost and KV-cache burden of full softmax attention while retaining much of the quality of Transformer models. A practical way to obtain such models is to convert a pretrained Transformer instead of pretraining a new architecture from scratch, but this conversion is still brittle. Simply copying the teacher attention projections into a Gated DeltaNet (GDN) student does not specify the new recurrent decay, write, and output-gating dynamics. As a result, the converted model often starts in a poor dynamical regime and must spend many distillation tokens repairing initialization rather than learning the remaining teacher behavior. We propose Taylor-Calibrate, a lightweight initialization method for hybrid GDN students. The method uses Taylor-guided teacher attention statistics to set the value projection, memory timescale, write gates, and output gate, then applies a short per-layer alignment step to match each converted layer to the teacher output. Across four teacher settings and three retained-layer policies, Taylor-Calibrate gives substantially stronger zero-shot students, with up to an 88x improvement in a representative ablation, and reaches matched recovery targets with 4.9x–9.2x fewer training tokens than naive conversion.

22.
arXiv (CS.AI) 2026-06-24

AutoSpec: Safety Rule Evolution for LLM Agents via Inductive Logic Programming

arXiv:2606.24245v1 Announce Type: cross Abstract: Large language model (LLM) agents increasingly automate complex tasks by integrating language models with external tools and environments. However, their autonomy poses significant safety risks: agents may execute destructive commands, leak sensitive data, or violate domain constraints. Existing safety approaches face a fundamental tradeoff: hand-crafted rules are interpretable but brittle, with overly conservative rules blocking safe operations (high false positives) while permissive rules miss unsafe behaviors (high false negatives). Neural classifiers lack the interpretability required for safety-critical deployments. We present AutoSpec, a framework that automatically evolves deployed expert-designed safety rules from user safe/unsafe annotations through counterexample-guided inductive synthesis (CEGIS) guided by inductive logic programming (ILP). Starting from the expert rules and a stream of annotated traces, AutoSpec iteratively evaluates rules, mines false-positive and false-negative counterexamples, uses ILP to learn which predicates discriminate them, generates candidate rule edits, and verifies candidates to select the best revision. The key insight is that ILP efficiently identifies predicates that appear frequently in false negatives but rarely in false positives (or vice versa), dramatically pruning the exponential search space of rule edits. This continues until convergence, producing interpretable rules that balance precision and recall. We evaluate AutoSpec on 291 execution traces spanning code execution and embodied agent domains. AutoSpec raises rule F1 to 0.98 and 0.93 across the two domains, achieving up to 94% false positive reduction while maintaining high recall, and converges within 4-5 iterations. The ILP-guided approach achieves up to 4.8x higher F1 than heuristic CEGIS. The learned rules are human-readable, auditable, and generalize to unseen scenarios.

23.
arXiv (CS.AI) 2026-06-16

User as Code: Executable Memory for Personalized Agents

作者:

arXiv:2606.16707v1 Announce Type: new Abstract: A personalized AI agent needs a user memory: a persistent model of who the user is, built across many conversations and consulted on each new one. Today this memory is almost always stored as unstructured text, a knowledge graph, or a flat store of facts, and consulted by retrieval – fetching the entries most similar to the current request. Such "bag-of-facts" memory recalls individual facts well, but because storing a fact and acting on it are separate steps, it struggles to resolve contradictions, aggregate over many records, or enforce rules. We argue that user memory should instead be executable. We introduce User as Code (UaC), a paradigm in which an agent's model of a user is a living software project: typed Python objects hold the user's state and ordinary Python functions encode the rules that govern it, so representing and reasoning about the user happen in one medium an interpreter can run. The enabling mechanism is a two-phase pipeline: an append-only log that never discards a fact, periodically checkpointed into typed code. This changes what memory can do. On standard long-term conversation benchmarks, UaC matches both a full-context upper bound and the strongest prior memory systems on recall (78.8% on LOCOMO). Its advantage emerges where representation matters most. On aggregate questions over a user's history – "how many international trips did I take last year?" – retrieval-based memory collapses (6-43%) while UaC stays near-perfect (99%), because the answer is a one-line computation over typed state rather than a search over text. And because its rules execute deterministically whenever the state changes, UaC can surface unsolicited, safety-critical alerts – such as a newly prescribed drug that conflicts with an allergy recorded months earlier – a capability query-driven memory cannot provide.

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

Diffusion Transformer World-Action Model for AV Scene Prediction

Action-conditioned world models let an autonomous vehicle predict future camera scenes from its own planned controls, enabling planning and simulation without real-world rollouts, but at compact, trainable scale the futures are ambiguous and the field's standard distortion metrics actively mislead: they reward a blurry regression mean over a realistic prediction. We confront this with a compact latent world model that, given the present front-camera latent and a sequence of ego-actions, predicts future scene latents a frozen decoder renders to $256 \times 256$ frames up to 8 seconds ahead, evaluated on 150 held-out nuScenes scenes. We first benchmark where to predict: across six frozen encoders spanning four representation families, V-JEPA2 with temporal context reduces steering RMSE by 40% over the best single-frame encoder. We then train a latent Diffusion Transformer (DiT) and, through a controlled diagnosis, identify the four ingredients it needs: spatial tokens, the $x_0$ objective, residual anchoring, and sampling matched to target uncertainty. In a Stable-Diffusion-VAE encode-predict-decode pipeline we expose the central tension: distortion metrics (cosine similarity, SSIM) favor the blurry mean, masking that the diffusion model is far closer to the real frame distribution. Inception-based FID and KID reveal a clean perception-distortion frontier: diffusion attains KID 0.078 versus 0.375 for regression ($4.8\times$ better), and a deployable train-derived calibration makes this practical without test-time ground truth. The model is genuinely action-controllable (steering drives scene displacement, Spearman $\rho = 0.81$, vs $-0.18$ for regression). We trace limited single-pass motion to a shared-present anchor and engineer a compact 1.7M-parameter "jump" model that recovers full ground-truth motion magnitude ($1.02\times$ GT), where single-pass models capture less than half.

25.
arXiv (CS.CL) 2026-06-19

Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias

LLM-as-a-Judge has become the dominant evaluation paradigm for language models, but judge validation in practice relies on exact-match agreement, a metric that does not correct for chance and systematically overstates discriminative ability. We present the largest systematic evaluation of LLM-as-a-Judge to date: 21 judges from nine providers across MT-Bench, JudgeBench, and RewardBench, evaluated under three protocols (agreement, consistency, bias audit) over 118 runs and approximately 541,000 individual judgments. Four findings emerge, consistent across the full cohort, including the April 2026 frontier: kappa deflation between exact match and Cohen's kappa is universal (33–41 pp on MT-Bench), judge rankings shift by up to 14 positions across benchmarks, high test–retest reliability (>0.95) coexists with severe position bias (>0.10) in two production-deployed judges (instantiating a consistency–bias paradox), and verbosity bias is small (