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

Sensitivity Shaping for Latent Modeling

arXiv:2606.14585v1 Announce Type: cross Abstract: Generative dynamics models enable planning in challenging robotic systems, but safe deployment requires reliably detecting policy-induced out-of-distribution (OOD) transitions. Existing methods typically treat the learned dynamics as fixed and attach post hoc support surrogates. We show that these surrogates can fail when the dynamics are locally insensitive to critical action choices: unsupported control actions may produce latent predictions that resemble demonstrated transitions, suppressing OOD signals despite large true predictive errors. To address this, we introduce support-conditioned control-sensitivity regularization, which promotes sensitive local response to control input changes in learned dynamics in high-support training regions. This preserves control-induced variation while limiting unstable extrapolation due to weak empirical support. Experiments in vision-based obstacle avoidance, manipulation, and real-robot navigation show improved OOD detection and safer closed-loop planning.

02.
arXiv (CS.AI) 2026-06-12

SAIGuard: Communication-State Simulation for Proactive Defense of LLM Multi-Agent Systems

arXiv:2606.12474v1 Announce Type: cross Abstract: LLM-based multi-agent systems (MAS) solve complex tasks through inter-agent collaboration, but their communication-driven nature also allows security risks to spread across agents and trigger system-wide failures. Existing MAS defenses mainly follow a reactive paradigm after execution by detecting and isolating harmful agents, which may cause irreversible damage and degrade collaborative utility. To address this, we propose a proactive defense framework for MAS security, namely a Simulation-aware Interception Guard (SAIGuard). SAIGuard performs communication-state simulation over the MAS interaction graph, estimates the impact of incoming messages on local agent states and the global MAS state, and detects risky messages via reconstruction deviations from benign communication patterns. Instead of isolating agents, SAIGuard sanitizes or regenerates suspicious messages before it propagation into system. Experiments across diverse topologies and attack scenarios show that SAIGuard reduces attack success rates while maintaining MAS utility, outperforming reactive defenses.

03.
arXiv (CS.AI) 2026-06-24

JupOtter: Cell-Level Bug Detection in Jupyter Notebooks

arXiv:2606.23877v1 Announce Type: cross Abstract: Jupyter Notebooks are an increasingly popular coding environment used across many domains, especially in Python-based data science and scientific computing. Originally used for prototyping and interactive exploration, notebooks are increasingly used to develop more complex programs, leading to a rapid rise in buggy notebooks on platforms like GitHub. To address this trend, we present JupOtter, a bug detection system designed specifically for Jupyter Notebooks. JupOtter features three novel contributions: (1) a notebook-specific tokenization strategy that preserves cell structure, (2) a cell-level bug prediction technique, and (3) a new labeled dataset, OtterDataset, containing over 21,000 notebooks annotated for fine-grained cell-level bug detection. JupOtter achieves cell-level bug detection F1 scores that surpass static analyzers and large language models in two out of three evaluation datasets.

04.
arXiv (CS.AI) 2026-06-16

Resilient Consensus in Agentic AI

arXiv:2606.15024v1 Announce Type: cross Abstract: Large language model (LLM) agents are increasingly deployed in multi-agent systems where they must coordinate and agree on shared decisions. We ask whether classical resilient consensus theory, developed for deterministic agents, transfers to LLM agents that may behave adversarially. Framing LLM agreement as a Byzantine consensus game, we run controlled experiments on complete and general communication graphs. We find that prompted LLM agents fail to reach agreement that is achievable in principle: consensus can fail even in settings where classical theory guarantees that a convergent algorithm exists, and this failure persists across temperatures and horizons. At the same time, wrapping the agents with classical resilient consensus filters improves agreement. The benefit of filtering depends on how much robustness the underlying topology already provides. Our results suggest that classical resilient consensus theory is a useful lens for the safety of agentic AI.

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

Auteur: Language-Driven Cinematographic Framing for Human-Centric Video Generation

Generative video models have achieved remarkable visual fidelity and temporal coherence, yet intentional camera control remains elusive. Existing frameworks treat camera motion as a byproduct of pixel synthesis, producing trajectories that are stochastic, spatially inconsistent, and indifferent to the human subject driving the scene. In this work, we present Auteur, a method for language-driven, human-centric camera framing in generative video. Our core insight is that professional filmmakers conceive shots not as world-space trajectories but as framings defined relative to the actor, encoding shot size, angle, and composition as functions of human pose and motion. We formalize this intuition as a human-centric camera parameterization and introduce a Domain-Specific Language (DSL) that is convertible to standard 6-DoF camera parameters. A fine-tuned multimodal large language model then acts as a virtual director, mapping natural language descriptions and coarse human motion to sparse DSL keyframes that are deterministically interpolated into continuous camera trajectories, which are then provided as input to video generators. We train and evaluate Auteur on a new dataset of 34K aligned text, human motion, and DSL-annotated camera trajectories drawn from procedural synthesis and real-world movie footage from the CondensedMovies dataset. Auteur enables cinematographic framing of human-centered scenes, a capability largely absent in prior generative models. To assess this behavior, we propose new framing-focused metrics, and our experiments show that Auteur consistently outperforms existing methods. Project page is https://cyberiada.github.io/Auteur/

06.
arXiv (CS.LG) 2026-06-15

Classification of Astronomical Spectra Using PCA-Compressed Flux and Inverse-Variance Features

arXiv:2606.13978v1 Announce Type: cross Abstract: This paper evaluates a signal-processing and supervised-learning pipeline for classifying SDSS DR17 astronomical spectra into stars, galaxies, and quasars. Each spectrum is represented by its measured flux and inverse-variance information, combining spectral shape with a wavelength-dependent reliability profile. After resampling onto a common logarithmic wavelength grid, the flux and inverse-variance vectors are standardized and separately compressed using principal component analysis. The resulting components are concatenated and used to train several classifiers. The best performance was obtained with the LightGBM gradient-boosting classifier, reaching $94.6\%$ accuracy and $92.1\%$ balanced accuracy on the test set.

07.
arXiv (CS.LG) 2026-06-16

FlowRL: A Taxonomy and Modular Framework for Reinforcement Learning with Diffusion Policies

arXiv:2603.27450v2 Announce Type: replace Abstract: Thanks to their remarkable flexibility, diffusion models and flow models have emerged as promising candidates for policy representation. However, efficient reinforcement learning (RL) upon these policies remains a challenge due to the lack of explicit log-probabilities for vanilla policy gradient estimators. While numerous attempts have been proposed to address this, the field lacks a unified perspective to reconcile these seemingly disparate methods, thus hampering ongoing development. In this paper, we bridge this gap by introducing a comprehensive taxonomy for RL algorithms with diffusion/flow policies. To support reproducibility and agile prototyping, we introduce a modular, JAX-based open-source codebase that leverages JIT-compilation for high-throughput training. Finally, we provide systematic and standardized benchmarks across Gym-Locomotion, DeepMind Control Suite, and IsaacLab, offering a rigorous side-by-side comparison of diffusion-based methods and guidance for practitioners to choose proper algorithms based on the application. Our work establishes a clear foundation for understanding and algorithm design, a high-efficiency toolkit for future research in the field, and an algorithmic guideline for practitioners in generative models and robotics. Our code is available at https://github.com/typoverflow/flow-rl.

09.
arXiv (CS.AI) 2026-06-17

Lean4Agent: Formal Modeling and Verification for Agent Workflow and Trajectory

arXiv:2606.06523v2 Announce Type: replace Abstract: Equipping Large Language Models (LLMs) to execute reliable multi-step workflows has become a central challenge in artificial intelligence. Despite recent advances in LLMs' agentic capabilities, most agent systems still lack formal methods for specifying, verifying, and debugging their workflow and execution trajectories. This challenge mirrors a long-standing problem in mathematics, where the ambiguity of natural languages (NLs) motivates the development of formal languages (FLs). Inspired by this paradigm, we propose **Lean4Agent**, to the best of our knowledge, the first framework that uses Lean4, a dependent-type FL to model and verify agent behavior. **Lean4Agent** launches **FormalAgentLib**, an extensible Lean4 library for formally modeling and verifying agent workflows' semantic consistency under explicit assumptions, and enabling localization of execution-time failures revealed by trajectories. Building on **FormalAgentLib**, we further develop **LeanEvolve**, which applies results in **FormalAgentLib** to revise workflows to enhance its capability. Extensive experiments on a hard problem subset of SWE-Bench-Verified and a subset of ELAIP-Bench across 5 leading LLMs indicate that the verification-passing workflows outperform the failing ones by an average of **11.94%**, and **LeanEvolve** further improves SWE performance by **7.47%** on average. Furthermore, **Lean4Agent** establishes a foundation for a new field of using expressive dependent-type FL to formally model and verify agent behavior.

10.
arXiv (CS.AI) 2026-06-12

Speculative Rollback Correction for Quality-Diverse Web Agent Imitation

arXiv:2606.12485v1 Announce Type: cross Abstract: Training interactive web agents through imitation learning from expert trajectories has emerged as a highly effective approach. However, determining the optimal timing for expert intervention presents a critical challenge in this context. Delayed intervention often leads to the accumulation of early-stage errors, pushing the page state into an irrecoverable regime. Conversely, premature or excessive intervention causes the agent to become overly reliant on expert policies, trapping the model in local optima characterized by a single, rigid trajectory. We propose Speculative Rollback Correction (SRC), a branch-level imitation framework for resettable agent environments. Instead of requesting teacher labels at every visited state or correcting only after a completed trajectory, SRC uses fixed-horizon branch review: the student executes a short speculative segment before teacher review, and the teacher localizes the first harmful deviation only when local progress breaks. Rollback preserves useful prefixes, while successful rollouts are filtered by a hard verifier and retained in a lightweight quality-diversity archive. The resulting data supports next-action supervised fine-tuning on both localized corrections and verifier-passing trajectories. On WebArena-Infinity, SRC collects 977 verifier-passing trajectories and 9,183 next-action examples; fixed-horizon review improves the recovery-versus-query tradeoff over step-level review while retaining verifier-passing solution variants. Code is available at https://github.com/LongkunHao/SRC_gui_agent.

11.
medRxiv (Medicine) 2026-06-16

Infections and suicide and self-harm: a population-based matched cohort study

Background Infections have been associated with adverse mental health outcomes, including suicide, but evidence beyond severe or central nervous system infections is limited. We investigated associations between a range of acute infections and subsequent suicide/self-harm outcomes. Methods We conducted six infection-specific matched cohort studies using English primary care records from the Clinical Practice Research Datalink Aurum (2007-2024), linked to hospital admissions and mortality data. Adults ([≥]18 years) with a primary care record of infection (gastroenteritis, lower respiratory tract [LRTI], skin/soft-tissue [SSTI], urinary tract [UTI], sepsis, meningitis/encephalitis [positive control]) were matched (age, sex, practice, calendar period) to up to five comparators without infection. We estimated hazard ratios (HRs) for suicide/self-harm outcomes using Cox regression, stratified by matched set and implicitly adjusting for matching factors, with additional adjustment for deprivation, lifestyle factors, and comorbidities. We examined whether associations varied over time, by infection severity, antimicrobial treatment, sex, and prior mental health conditions. Findings Cohorts ranged from 18,192 individuals with meningitis/encephalitis (matched to 90,915 without) to 398,099 with SSTI (matched to 1,743,747). After adjustment, individuals with infection had a higher hazard of suicide/self-harm outcomes than comparators across all cohorts: sepsis (HR 1.79, 95% CI 1.65-1.93), gastroenteritis (1.62, 1.55-1.70), meningitis/encephalitis (1.56, 1.32-1.84), UTI (1.41, 1.33-1.50), SSTI (1.37, 1.31-1.43), and LRTI (1.37, 1.31-1.44). Risk was highest in the year post-infection, attenuating over time, and was higher among severe infections and those without prior mental health conditions. Interpretation Common acute infections recorded in primary care are associated with increased risk of suicide and self-harm, particularly following severe infections and in the year post-infection. Findings support suicide risk monitoring following acute infection, particularly among individuals without prior mental health conditions, and highlight infection prevention as a potentially modifiable strategy in vulnerable populations. Funding Wellcome and La Caixa. Copyright This work is licensed under a Creative Commons Attribution (CC BY) licence.

12.
bioRxiv (Bioinfo) 2026-06-19

Simulation-based Bayesian deep learning enables uncertainty-aware tumor fraction estimation in cell-free DNA

Background: Estimating tumor fraction from whole-genome cell-free DNA sequencing is critical for liquid biopsy, but is hampered by weak signals and baseline noise at low tumor fractions. Existing computational methods often require matched controls or large labeled datasets for training and lack uncertainty quantification. To address these gaps, we developed purNPE, a Bayesian deep-learning framework trained without labeled cancer cell-free DNA samples. Specifically, purNPE leverages a two-part generative model: one component simulates diverse tumor copy-number profiles based on evolutionary genealogies, while a second, data-driven component learns and replicates realistic sequencing background patterns from cancer-free cell-free DNA. By training a Neural Posterior Estimator on synthetic tumor profiles augmented with learned noise, purNPE performs amortized inference in milliseconds without needing a reference sample set at inference. Results: In a real-world pan-cancer cohort, purNPE achieved comparable performance with existing methods against orthogonal mutant-allele-fraction validation (MAE = 0.066). In silico and semi-synthetic experiments suggested analytical sensitivity around 1% tumor fraction under the evaluated conditions and showed strong classification accuracy in low tumor fractions (AUC = 0.98 for TF [≤] 3% versus controls). Conclusions: This work provides a framework for using simulation-based inference to derive calibrated, uncertainty-aware TF estimates, offering a potential alternative to traditional data-dependent methods.

13.
arXiv (CS.AI) 2026-06-18

Caring Without Feeling: Affective Dynamics as the Control Layer of Human-AI Agent Collaboration

arXiv:2606.18259v1 Announce Type: cross Abstract: AI agents that plan, retain memory across sessions, invoke external tools and act with partial autonomy are transforming human–AI collaboration. Research on affective computing, simulated empathy in large language models, trust in automation and AI safety has illuminated important design principles, yet these literatures remain fragmented. No integrated account explains how affective cues operate within agentic collaboration – settings in which humans delegate, monitor and correct consequential tasks. This Review synthesises computational and interactional mechanisms of affective dynamics: the processes through which affective cues, emotion-like behaviour and perceived agent affect shape trust calibration, delegation decisions, error correction, dependence and governance. We trace how model-generated affective signals enter interaction loops that govern reliance, repair and oversight, and propose a framework that treats affect not as an internal property of AI but as a coordination layer through which humans and agents negotiate capability, uncertainty and responsibility. The framework provides a foundation for calibrated measurement, purposeful design and informed governance.

14.
arXiv (CS.CL) 2026-06-11

External Experience Serving in Production LLM Systems: A Deployment-Oriented Study of Quality-Cost Trade-offs

Production LLM systems accumulate reusable operational experience, but the practical deployment issue is not merely whether such experience can help. It is how different serving strategies trade off quality against online cost under realistic constraints. Injecting external experience can improve task quality, yet it also increases prompt burden, latency, and serving pressure. We study external experience serving as a deployment-oriented quality-cost trade-off problem. We evaluate this question in a real production moderation setting, with tool-use and GPQA as supporting contrast tasks that expose different output-cost regimes. We compare no-experience baselines, random experience controls, global prompt injection, and retrieval-based selective injection, and analyze both task quality and serving cost. The results show that, once experience becomes case-dependent, selective retrieval provides a stronger operating point than unconditional global injection. They further show that retrieval quality matters more than simply increasing Top-$K$, and that the same serving policy can exhibit substantially different cost-benefit profiles across short-output and decode-heavy regimes. These findings suggest that external experience is best treated as a selective, cost-aware serving decision rather than as a universal add-on. Overall, in the settings studied here, external experience pays off only when both the serving interface and the task-specific cost structure make its quality gains worth the online cost.

16.
arXiv (quant-ph) 2026-06-19

Passive-User Bell-State Loop-Back Key Establishment without Quantum Detectors at the User Nodes

arXiv:2606.19551v1 Announce Type: new Abstract: We propose and analyze a Bell-state extension of the Loop-Back quantum key distribution architecture for secret-key establishment between two passive users that do not require quantum transmitters or quantum detectors. In the proposed setting, a single active station, Alice, provides the entangled-state infrastructure, retains one qubit of an initially prepared Bell pair, and sends the traveling subsystem through two passive users, denoted by $B_1$ and $B_2$. Each passive user applies a local Pauli operation to the same traveling subsystem, so that the operation observed by Alice is only the effective composition $U_{\mathrm{eff}}=U_2U_1$. After the subsystem returns, Alice performs a Bell-state measurement and, using her private knowledge of the initial Bell state, deterministically identifies the effective Pauli operation. However, the individual factors $U_1$ and $U_2$ remain algebraically hidden from Alice whenever the local choices are uniformly and independently selected. The public effective operation acts as a parity-like constraint: each passive user can infer the operation applied by the other from its own private choice, while the active station learns only the global composition. This construction transfers the essential distributed-transformation mechanism of passive-user Loop-Back QKD to the entangled-state regime. Unlike single-qubit passive-user schemes, whose useful events are intrinsically post-selected, the Bell-state version is limited primarily by the success probability of the Bell-state measurement. We discuss the algebraic structure of the protocol, its interpretation as an infrastructure-assisted mediated key-establishment mechanism, and the physical assumptions required to protect passive Pauli modulators against active injection or Trojan-horse-type attacks.

17.
arXiv (CS.LG) 2026-06-19

Multi-Modal Contrastive Learning for Implicit Earth Embeddings via Location Tying

arXiv:2606.20167v1 Announce Type: new Abstract: Spatial prediction tasks are often limited by a lack of high-quality labelled ground-truth observations. To overcome this challenge, self-supervised pre-training is a possible solution, with contrastive learning dominant for location encoders. Those approaches usually align geographic coordinates with just one additional modality. We propose two multimodal contrastive learning architectures: Multimodal Embedding via Location Tying (MELT) and Sequential Alternating Location Training (SALT). These architectures expand this framework beyond two modalities by utilising unpaired geospatial data. Both methods are technically viable and match the performance of the strongest two-modality baseline (SATCLIP) across four downstream tasks. However, increasing the number of modalities does not consistently improve performance, suggesting that the chosen location encoder is the main limitation - the contrastive objective reaches its peak early, regardless of modality diversity or pre-training volume. MELT provides more stable training than SALT and presents a stronger foundation for future scaling.

18.
arXiv (math.PR) 2026-06-24

Gradient Mean-Field Dynamics with Measure-Valued States: Well-Posedness, Chaos, and Long-Time Stability

arXiv:2606.24385v1 Announce Type: new Abstract: We study a stochastic mean-field interacting particle system whose state space is $\Y = \Tt^d \times \cP(U)$, where the first component represents a spatial variable and the second one is a probability measure over a compact metric space $U$. The dynamics are driven by locally Lipschitz drift operators: the spatial component evolves according to a Brownian diffusion, while the measure-valued component is perturbed by a projected cylindrical noise acting in the Arens–Eells space. We first establish existence and uniqueness of strong solutions for both the $N$-particle system and the associated nonlinear McKean–Vlasov equation under locally Lipschitz and linear growth assumptions on the drift coefficients. We then prove propagation of chaos: as $N\to\infty$, the empirical measure converges in expectation in Wasserstein–1 distance towards the unique McKean–Vlasov solution. Further, we investigate exponential convergence of the nonlinear McKean–Vlasov dynamics towards a unique invariant measure.

19.
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.

20.
arXiv (CS.AI) 2026-06-15

When Sample Selection Bias Precipitates Model Collapse

arXiv:2606.13732v1 Announce Type: new Abstract: The proliferation of recursive training on synthetic data can alleviate data scarcity but risks model collapse, where repeated training erodes distributional tails and homogenizes outputs. Data selection is widely viewed as a remedy, yet its reliability depends critically on the reference distribution used by the verifier. We show that in low-resource verification regimes, where each verifier observes only a small, fragmented, and biased slice of the target manifold, selection itself becomes biased. This situation naturally arises in low-resource data silos such as healthcare consortia or proprietary financial institutions, where raw data cannot be pooled and local references are inherently incomplete. As a result, selection preferentially retains samples aligned with the local manifold while pruning globally relevant tail modes, turning from a safeguard against collapse into a mechanism that precipitates it. We theoretically prove that such siloed selection accelerates collapse and induces power-law diversity decay. As an initial mitigation, we construct Wasserstein proxy references from multiple silos without sharing raw data. Empirical results confirm that local-reference selection fails on skewed distributions, whereas collaborative proxy references mitigate diversity degradation, suggesting that recursive synthetic-data pipelines require particular caution when real-data coverage is fragmented or scarce.

21.
arXiv (CS.CL) 2026-06-19

Omnilingual SONAR: Cross-Lingual and Cross-Modal Sentence Embeddings Bridging Massively Multilingual Text and Speech

Cross-lingual sentence encoders typically cover only a few hundred languages and often trade downstream quality for stronger alignment, limiting their adoption. We introduce OmniSONAR, a new family of omnilingual, cross-lingual and cross-modal sentence embedding models that natively embed text, speech, code, and mathematical expressions in a single semantic space, while delivering state-of-the-art downstream performance at the scale of thousands of languages, from high-resource to extremely low-resource varieties. To reach this scale without representation collapse, we use progressive training. We first learn a strong foundational space for 200 languages with an LLM-initialized encoder-decoder, combining token-level decoding with a novel split-softmax contrastive loss and synthetic hard negatives. Building on this foundation, we expand to several thousands language varieties via a two-stage teacher-student encoder distillation framework. Finally, we demonstrate the cross-modal extensibility of this space by seamlessly mapping 177 spoken languages into it. OmniSONAR halves cross-lingual similarity search error on the 200-language FLORES dataset and reduces error by a factor of 15 on the 1,560-language BIBLE benchmark. It also enables strong translation, outperforming NLLB-3B on multilingual benchmarks and exceeding prior models (including much larger LLMs) by 15 chrF++ points on 1,560 languages into English BIBLE translation. OmniSONAR also performs strongly on MTEB and XLCoST. For speech, OmniSONAR achieves a 43% lower similarity-search error and reaches 97% of SeamlessM4T speech-to-text quality, despite being zero-shot for translation (trained only on ASR data). Finally, by training an encoder-decoder LM, Spectrum, exclusively on English text processing OmniSONAR embedding sequences, we unlock high-performance transfer to thousands of languages and speech for complex downstream tasks.

22.
arXiv (CS.CL) 2026-06-15

Fodor and Pylyshyn's Systematicity Challenge Still Stands

The recent successes of neural networks producing human-like language have caused significant stir in cognitive science, with many researchers arguing that classical puzzles about human cognition and challenges to artificial intelligence are being solved by neural networks. A notable case is the argument from systematicity due to Jerry Fodor and Zenon Pylyshyn, argues that humans display systematic biconditional dependencies. For example, someone can understand the sentence "John saw Mary" just in case that they understand the sentence "Mary saw John." Symbolic systems explain this systematicity of language and thought, while neural networks offer no immediate explanation. Several recent articles argue that this challenge has now been met by neural networks. In particular, Brenden Lake and Marco Baroni argue that their meta-learning for compositionality protocol matches and perhaps explains human systematicity. We demonstrate that these conclusions are premature. Among other results, we found that their model struggles to learn rules that are even slightly out of distribution compared to their training data. Furthermore, the model behaves unsystematically even on many within-distribution problems. We conclude that Fodor and Pylyshyn's challenge to neural networks remains unmet.

23.
medRxiv (Medicine) 2026-06-22

Demographic Calibration Gaps in Breast Cancer Risk Prediction: Introducing the Demographic Calibration Gap Score

作者:

ABSTRACT: Most breast cancer prediction studies skip calibration reporting entirely. Fewer still examine calibration by demographic subgroup. Predicted probabilities that are systematically off for specific racial or gender groups produce biased clinical decisions, and aggregate statistics will not catch that. Objective: To introduce the Demographic Calibration Gap Score (DCGS), a metric that measures how much calibration error varies across demographic subgroups, and to show how it performs across five classifiers, four calibration conditions, and two datasets. Methods: Five classifiers were trained on the Wisconsin Diagnostic Breast Cancer dataset (n=569) and evaluated on a breast cancer cohort from MIMIC-IV (n=1,316). Three global calibration methods were applied: no calibration, Platt scaling, and isotonic regression. A fourth condition, subgroup-targeted Platt scaling, was applied to the MIMIC cohort. DCGS was computed as across racial and gender subgroups, with 95% bootstrap confidence intervals. Conformal prediction coverage and Demographic Coverage Gap (DCG) were reported. Results: On Wisconsin, all five models achieved AUROC above 0.98 and ECE below 0.12. Performance fell sharply on the MIMIC external cohort: AUROC dropped to 0.45-0.57 for base and globally calibrated variants, confirming distributional shift. DCGS exceeded the 0.05 clinical significance threshold in 28 of 40 model-calibration combinations on the race axis. Neither global Platt nor isotonic calibration reliably reduced DCGS below that threshold. Conformal coverage collapsed to roughly 25% on MIMIC, and racial DCG exceeded 0.15 for all 20 model-variant combinations. Conclusions: Reducing population-level ECE through global recalibration does not reliably close demographic calibration gaps. DCGS gives researchers a direct, standardized way to detect and report those disparities. Code and the DCGS computation library are released as open-source Python under the MIT License.

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

Iterative Visual Thinking: Teaching Vision-Language Models Spatial Self-Correction through Visual Feedback

Vision-language models (VLMs) achieve strong singleshot spatial grounding, yet lack any mechanism to observe and correct their own predictions. We find that naively prompting a VLM to iterate over rendered visualizations of its predictions causes catastrophic failure: Acc@0.5 on referring expression comprehension collapses from 79.6% to 48.7% (a 31 percentage point drop), revealing a fundamental gap between grounding capability and self-correction ability. We propose Iterative Visual Thinking (IVT), a closed-loop framework in which the model predicts a bounding box, observes the prediction rendered on the image, and iteratively refines through visual feedback. A two-phase training recipe closes the self-correction gap: first, we exploit the base model's own predictions as realistic errors and prompt a teacher VLM to generate corrective reasoning traces, yielding supervised data without human annotation; second, we apply Group Relative Policy Optimization (GRPO) with a simple IoU reward to stabilize multi-step refinement. On a mixed benchmark spanning RefCOCOg, Ref-Adv, and Ref-L4 (505 test samples), SFT warm-up with IVT surpasses the single-shot base model on every metric: Acc@0.5 rises to 82.0% (+2.4pp), Acc@0.7 to 74.1% (+3.2pp), and Acc@0.9 to 48.3% (+2.8pp). GRPO further reduces per-step IoU degradation by 5x, stabilizing the refinement trajectory. All training uses only 2,400 samples on a single GPU, demonstrating that spatial self-correction is a learnable capability that can be instilled at modest scale.

25.
arXiv (CS.CL) 2026-06-12

LingxiDiagBench: A Multi-Agent Framework for Benchmarking LLMs in Chinese Psychiatric Consultation and Diagnosis

Mental disorders are highly prevalent worldwide, but the shortage of psychiatrists and the inherent subjectivity of interview-based diagnosis create substantial barriers to timely and consistent mental-health assessment. Progress in AI-assisted psychiatric diagnosis is constrained by the absence of benchmarks that simultaneously provide realistic patient simulation, clinician-verified diagnostic labels, and support for dynamic multi-turn consultation. We present LingxiDiagBench, a large-scale multi-agent benchmark that evaluates LLMs on both static diagnostic inference and dynamic multi-turn psychiatric consultation in Chinese. At its core is LingxiDiag-16K, a dataset of 16,000 EMR-aligned synthetic consultation dialogues designed to reproduce real clinical demographic and diagnostic distributions across 12 ICD-10 psychiatric categories. Through extensive experiments across state-of-the-art LLMs, we establish key findings: (1) although LLMs achieve high accuracy on binary depression–anxiety classification (up to 92.3%), performance deteriorates substantially for depression–anxiety comorbidity recognition (43.0%) and 12-way differential diagnosis (28.5%); (2) dynamic consultation often underperforms static evaluation, indicating that ineffective information-gathering strategies significantly impair downstream diagnostic reasoning; (3) consultation quality assessed by LLM-as-a-Judge shows only moderate correlation with diagnostic accuracy, suggesting that well-structured questioning alone does not ensure correct diagnostic decisions. We release LingxiDiag-16K and the full evaluation framework to support reproducible research at https://github.com/Lingxi-mental-health/LingxiDiagBench.