# Research Papers — 2026-06 · bundle page 79 of 260 > Full-text markdown of papers #3901–3950 of 12993 published in 2026-06, newest first. AI-readable mirrors of the full 12993-paper 2026-06 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-06/page/79 [← Previous bundle](https://aipod.dev/papers/archive/2026-06/page/78.md) [Next bundle →](https://aipod.dev/papers/archive/2026-06/page/80.md) ### Black-Box Forensics for Conversational LLM Agents Authors: Isadora White, Yasaman Jafari, Taylor Berg-Kirkpatrick Published: 2026-06-21 Categories: cs.CR, cs.CL arXiv: https://arxiv.org/abs/2606.22698v1 As LLM-powered scams proliferate, black-box forensics for conversational LLM agents offers a path to accountability for systems hidden behind anonymous endpoints. Identifying the base model behind a chatbot endpoint (attribution), without model parameter access or knowledge of the hidden system prompt, would let investigators trace AI-enabled scams back to the providers whose models power them. Detecting when two endpoints run the exact same system prompt (fingerprinting), even one novel and unseen, would link individual scams into criminal networks and expose silent API changes. We conduct an empirical investigation of both capabilities. Our attribution classifiers identify the base model behind an agent with 98% accuracy from a few turns of non-adversarial conversation. Attribution of system prompts, while possible, requires retraining on a large amount of data for each prompt; system prompts in the wild are unbounded and ever-changing, making this approach costly. To tackle this more open-ended setting, our cross-encoder fingerprinting method achieves an AUC of 0.768 and an F1 of 0.703 on entirely unseen system prompts, and aggregating 50 interaction conversations from each target agent boosts AUC to 0.943. Conversational agents with unseen system prompts can thus be fingerprinted with robust accuracy from a few turns of ordinary conversation. --- ### NullFlow: One-Step Generative Reconstruction Authors: Xiao Shi, Edward P. Chandler, Chicago Y. Park, Shirin Shoushtari, Ulugbek S. Kamilov Published: 2026-06-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.22696v1 We propose NullFlow, a principled framework for one-step generative image reconstruction. Our key idea is to confine the generative flow to a measurement-consistent subspace. Because the flow never leaves this subspace, NullFlow needs no separate data-fidelity corrections, unlike existing solvers. NullFlow samples in a single network evaluation by learning the flow's average velocity, avoiding the step-by-step integration of traditional flow matching methods. We prove that the average velocity of this constrained flow yields a training objective whose global minimizer is a one-step posterior sampler. We show on image inpainting that NullFlow matches state-of-the-art diffusion solvers while cutting inference from hundreds of network evaluations to one. --- ### SATURN: Symbolic Spatial Reasoning for Multi-Perspective Grounding Authors: Danial Kamali, Tanawan Premsri, Shreya Rajpal, Amir Zadeh, Chuan Li, Parisa Kordjamshidi Published: 2026-06-21 Categories: cs.CV, cs.SC arXiv: https://arxiv.org/abs/2606.22694v2 Vision-Language Models (VLMs) remain unreliable when spatial reasoning requires composing relations whose meanings depend on frames of reference. Existing tool-augmented spatial reasoning methods make reasoning more explicit, but often rely on low-level geometric procedures and hard binary decisions over noisy perception. We propose SATURN, a neuro-symbolic framework for perspective-aware compositional spatial reasoning. SATURN reconstructs an approximate 3D scene, derives soft perspective-aware spatial predicates, and composes them with a training-free Pythonic symbolic executor, separating perception from reasoning while preserving uncertainty through multi-hop inference. We also introduce 3D FORCE, a diagnostic benchmark that controls reasoning depth, view, and perspective composition for spatial arrangement grounding (SAG) and referring expression grounding (REF). On 3D FORCE, VLMs and spatially trained models degrade sharply as depth and perspective complexity increase, whereas SATURN degrades the least and outperforms every baseline at each depth. On the real-world MindCube benchmark, SATURN achieves \(78.06\%\) overall accuracy, outperforming the strongest baseline by \(14\) percentage points. --- ### SATURN: Symbolic Spatial Reasoning for Multi-Perspective Grounding Authors: Danial Kamali, Tanawan Premsri, Shreya Rajpal, Amir Zadeh, Chuan Li, Parisa Kordjamshidi Published: 2026-06-21 Categories: cs.CV, cs.SC arXiv: https://arxiv.org/abs/2606.22694v1 Vision-Language Models (VLMs) remain unreliable when spatial reasoning requires composing relations whose meanings depend on frames of reference. Existing neuro-symbolic methods make reasoning more explicit, but often depend on brittle geometric procedures and hard decisions over noisy perception. We propose SATURN, a neuro-symbolic framework for perspective-aware compositional spatial reasoning. SATURN reconstructs an approximate 3D scene, derives soft perspective-aware spatial predicates, and composes them with a training-free Pythonic symbolic executor, separating perception from reasoning while preserving uncertainty through multi-hop inference. We also introduce 3D FORCE, a diagnostic benchmark that controls reasoning depth, view, and perspective composition across spatial arrangement grounding (SAG) and referring expression grounding (REF). On 3D FORCE, VLMs and spatially trained models degrade sharply as depth and perspective complexity increase, whereas SATURN remains stable and outperforms strong baselines. On the real-world MindCube benchmark, SATURN achieves 78.57% overall accuracy, outperforming the strongest baseline by 14 pp. --- ### VISTA Architect: A graph database-oriented health AI system demonstrated in multidisciplinary tumor boards Authors: Tuomo Kiiskinen, Jason Fries, Philip Adamson, David Wu, Timothy John Ellis-Caleo, Aaron Fanous, Balasubramanian Narasimhan, Joel Neal, Sylvia Plevritis, Manuel A. Rivas Published: 2026-06-21 Categories: cs.AI, cs.CL, cs.DB, cs.IR arXiv: https://arxiv.org/abs/2606.22692v1 We introduce VISTA Architect, a database-oriented AI architecture for integrating large language models (LLMs) with longitudinal electronic health records (EHRs). At ingestion, it transforms complex clinical documentation into a persistent, provenance-linked knowledge graph, eliminating repeated reprocessing of raw records at query time. The architecture has two layers: a source-faithful MEDS Graph preserving granular EHR structure with full provenance, and a clinically abstracted Timeline Object Architecture (TOA) that uses graph-guided LLM extraction to synthesize a concise timeline of deduplicated, temporally coherent clinical events. This addresses key limitations of direct long-context prompting and retrieval-augmented generation (RAG), which often miss temporal relationships and incur high cost and latency from repeated raw-text processing. By precomputing clinical synthesis once, downstream queries access an organized patient state and traverse to source documentation only when detailed verification is needed. We demonstrate the system in multidisciplinary thoracic oncology tumor boards at Stanford Medicine, where precise reconstruction of patient histories is critical. Across 1,180 patients, VISTA Architect achieved 96.4% accuracy (mean 9.75/10) on 15 tumor board-salient variables (17,700 evaluations; 95% CI 96.1-96.7%), surpassing a matched BM25 RAG baseline and recent benchmarks for LLM-based clinical extraction. An agentic interface reduced preparation for a 30-patient held-out cohort to about 2.2 minutes without sacrificing accuracy. While configured here for thoracic oncology, the modular design adapts to other specialties through customizable event definitions, episode structures, and agentic tools; validation beyond thoracic oncology remains future work. --- ### GARIP: A Running-Average Moving Reference for Last-Iterate Self-Play in Two-Player Zero-Sum Games Authors: Can Savcı Published: 2026-06-21 Categories: cs.MA, cs.GT, cs.LG arXiv: https://arxiv.org/abs/2606.22688v1 Self-play with naive gradient ascent cycles in two-player zero-sum games: the last iterate orbits the equilibrium. Modern methods restore last-iterate convergence by regularizing toward a reference policy -- MMD a fixed one (reaching only the regularized equilibrium), R-NaD a periodic snapshot (the engine of DeepNash). We study GARIP, which anchors to the running average, and isolate what the choice of reference controls. Our central result is a mechanism: collapse tracks the peak lag of the reference, and among causal convex averages of a fixed mean lag the running average (flat profile, peak $=$ mean) uniquely minimizes that peak, while a snapshot's sawtooth has peak $= 2\times$ mean (a one-line theorem). Two consequences follow. Convergence: we prove local last-iterate convergence at constant anchor strength -- the anchor scales the base map's rotation by $1-β$, crossing the stability boundary and turning a recurrent base into a contraction (global convergence is conjectured at small $β$; we characterize a large-$β$ consensus failure). Robustness: GARIP matches R-NaD's peak performance -- on matrix games, the Coin Game, and the board games Connect Four/Othello, both moving references are far more robust than fixed-magnet and magnet-free baselines -- but is the better hyperparameter default; we report it both ways: over the full grid collapse rates are statistically indistinguishable, yet at conventional parameterizations a matched-mean-lag setting collapses in 0/40 vs 10/40 seeds (a snapshot matches it only by knowing to shorten $K$). The boundaries: an anticipatory (negative-weight) reference does better still on the stale side, and the advantage appears only where naive self-play cycles (five deep self-play loops). All experiments are pure JAX and reproducible. --- ### The Geometry of Refusal: Linear Instability in Safety-Aligned LLMs Authors: Shivam Ratnakar, Kartikeya Vats Published: 2026-06-21 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2606.22686v2 Modern Large Language Models (LLMs) rely on extensive safety alignment, yet the mechanistic basis of refusal remains opaque. In this work, we investigate whether safety compliance is a deep semantic decision or a manipulable linear feature. We introduce Contrastive Logit Steering (CLS), a zero-optimization framework that isolates the "refusal direction" by contrasting hidden states derived from safe and unrestricted system prompts. Unlike representation engineering methods that intervene on internal activations, CLS operates directly on the output distribution, serving as a diagnostic probe for alignment fragility. When coupled with prefix injection to bypass initial refusal reflexes, this method induces a phase transition where guardrails collapse. Our experiments on 7 model families reveal that safety implementation is architecturally deterministic. While models like Llama-3.1 exhibit a "Late Decision" topology that is easily bypassed by CLS (reaching 95% ASR in approximately one second), others like Qwen-2.5 demonstrate "Early Divergence" by integrating safety mid-computation. Direct comparison with established activation-level steering methods shows that CLS achieves substantially higher attack success rates on Llama 2 (73% vs. 22.6%) and Qwen 7B (91% vs. 79.2%), demonstrating that logit-level intervention exposes alignment vulnerabilities that hidden-state methods underestimate. Beyond attacks, we show that this linearity enables bidirectional control: inverting the steering vector "hardens" models against jailbreaks without retraining. Our findings suggest that current alignment techniques create a steerable "safety axis" that serves as both a critical vulnerability and a precise primitive for defense. --- ### Only Ask What You Don't Know: Grounded Delta Planning for Efficient Multi-step RAG Authors: Wei-Chieh Chou, Xuanjun Chen, Jian-Ren Lin, Claire Lin, Hung-yi Lee, Jyh-Shing Roger Jang Published: 2026-06-21 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.22681v2 Multi-hop question answering remains challenging for Retrieval-Augmented Generation (RAG) because existing approaches either propagate errors across iterative retrieval rounds or over-generate reasoning steps, increasing cost without improving accuracy. We propose Grounded Delta Planning RAG (GDP-RAG), a plan-based framework that targets only the information delta based on three simple design choices: (1) preliminary retrieval to ground planning before execution, (2) a gap-conditioned planning prompt that asks only for missing information, and (3) a skeletal trajectory that pairs each subquery with a Thought capturing evidence from preliminary retrieval and carrying it through to the final answer. GDP-RAG focuses computation on unresolved gaps, yielding concise, reliable reasoning trajectories. Extensive experiments on HotpotQA, 2WikiMultiHopQA, and MuSiQue show that GDP-RAG achieves the highest accuracy (60.63%) among all compared systems while maintaining a cost-of-pass of 0.51, 22% lower than PAR-RAG (0.65) and 68% lower than KnowTrace (1.57), with no method achieving both higher accuracy and lower cost. --- ### Only Ask What You Don't Know: Grounded Delta Planning for Efficient Multi-step RAG Authors: Wei-Chieh Chou, Xuanjun Chen, Jian-Ren Lin, Claire Lin, Hung-yi Lee, Jyh-Shing Roger Jang Published: 2026-06-21 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.22681v1 Multi-hop question answering remains challenging for Retrieval-Augmented Generation (RAG) because existing approaches either propagate errors across iterative retrieval rounds or over-generate reasoning steps, increasing cost without improving accuracy. We propose Grounded Delta Planning RAG (GDP-RAG), a plan-based framework that targets only the information delta based on three simple design choices: (1) preliminary retrieval to ground planning before execution, (2) a gap-conditioned planning prompt that asks only for missing information, and (3) a skeletal trajectory that pairs each subquery with a Thought capturing evidence from preliminary retrieval and carrying it through to the final answer. GDP-RAG focuses computation on unresolved gaps, yielding concise, reliable reasoning trajectories. Extensive experiments on HotpotQA, 2WikiMultiHopQA, and MuSiQue show that GDP-RAG achieves the highest accuracy (60.63%) among all compared systems while maintaining a cost-of-pass of 0.51, 22% lower than PAR-RAG (0.65) and 68% lower than KnowTrace (1.57), with no method achieving both higher accuracy and lower cost. --- ### RigorBench: Benchmarking Engineering Process Discipline in Autonomous AI Coding Agents Authors: Meher Bhaskar Madiraju, Meher Sai Preetam Madiraju Published: 2026-06-21 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2606.22678v2 Agentic coding harnesses - such as Agent-Skills, Superpowers, and Agent-Rigor - are increasingly deployed to augment underlying LLMs for real-world software engineering tasks. Existing benchmarks evaluate these agents almost exclusively on outcome correctness: whether generated code passes tests or resolves issues. We argue that this outcome-only lens is insufficient: an agent that arrives at a correct solution through reckless trial-and-error, without planning, verification, or graceful recovery, is fundamentally less reliable than one that follows sound engineering discipline. We introduce RigorBench, the first benchmark designed to measure process discipline in AI coding agents. RigorBench evaluates these harnesses across five pillars: Planning Fidelity, Verification Coverage, Recovery Efficiency, Abstention Quality, and Atomic Transition Integrity. A composite RigorScore aggregates these dimensions into a single metric via a weighted sum. We curate a suite of 30 tasks spanning five categories - Plan-Then-Build, Verify-Or-Die, Doom Loop Gauntlet, Know When to Fold, and Don't Break the Build-and evaluate leading harnesses in a controlled with/without experimental design against baseline coding assistants. Our results show that structured process discipline not only improves process quality scores by an average of 41% but also raises downstream outcome correctness by 17%, providing the first quantitative evidence that how agents code matters as much as what they produce. We release the full benchmark, scoring rubrics, and trajectory analysis tools as open-source artifacts. --- ### Skin-Deep: A Geometric Diagnostic for Alignment Fragility in Large Language Model Representations Authors: Dongyub Jude Lee, Jungseob Lee, Seungyoon Lee, Seongtae Hong, Suhyune Son, Sugyeong Eo, Jaehyung Seo, Heuiseok Lim Published: 2026-06-21 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.22676v1 Alignment tuning is meant to make harmful-request refusal robust, yet this safety behavior can be erased by a small set of benign fine-tuning examples. This is a deployment risk for open-weight models because a checkpoint can pass refusal tests at release time and later lose refusal under low-cost downstream fine-tuning. Prior work has established these refusal failures, but existing studies do not show how to detect this fragility in the aligned model itself before an attack or fine-tuning intervention is run. We introduce Skin-Deep, a geometric diagnostic that detects alignment fragility directly from the aligned model's hidden-state activations before such an intervention is run and compresses the layer-wise safety geometry into a single scalar, the Geometric Fragility Score (GFS). Applied to twenty-one instruction-tuned models spanning six alignment recipes and 3B--32B parameters, Skin-Deep reveals a recurring low-rank safety subspace across model families. Direction ablations show that removing directions in this subspace weakens harmful-request refusal, providing causal evidence that the recovered geometry underlies refusal behavior. Crucially, GFS identifies, before any fine-tuning, the initially safe model that retains the most refusal after small-scale LoRA fine-tuning. These results establish GFS as a practical pre-deployment diagnostic for flagging fragile refusal behavior without running an attack. --- ### Data Evolution by Wittgenstein's Rule Following Authors: Aydin Ghojogh, Benyamin Ghojogh Published: 2026-06-21 Categories: stat.ML, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2606.22674v1 This paper introduces Wittgenstein's Rule Following (WRF) data evolution, a framework in philomatics for evolving or generating a new dataset from a sequence of previously observed datasets. The method is inspired by Ludwig Wittgenstein's rule-following considerations and his notion of family resemblance in Philosophical Investigations. Unlike standard synthetic data generation, where the goal is usually to sample from or augment a fixed distribution, WRF aims to continue the implicit rule expressed by a historical sequence of datasets while preserving resemblance to the previous datasets. WRF represents each dataset by structural descriptors rather than pointwise correspondences. These descriptors summarize geometric, distributional, clustering, and, in the supervised case, label-based properties of the data. The method predicts a rule-following target by extrapolating descriptor trajectories and a family-resemblance target by averaging historical descriptors. Candidate datasets are then generated from the observed history through balanced or bounded mixture recombination, scored according to these targets, and optionally refined through differentiable optimization in descriptor space. The proposed framework allows both sample size and feature dimension to vary over time and does not assume that the next dataset is a direct transformation of the last one. Simulations on synthetic and image datasets show that WRF can generate meaningful continuations of evolving datasets in both unsupervised and supervised settings. --- ### AgentLens: Interpretable Safety Steering via Mechanistic Subspaces for Multi-Turn Coding Agent Authors: Weidi Luo, Qiming Zhang, Yihao Quan, Mingyu Jin, Jie Cai, Chaowei Xiao, Jingcheng Niu, Zhen Xiang Published: 2026-06-21 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2606.22673v1 Coding agents based on large language models (LLMs) demonstrate remarkable autonomous capabilities, but they also introduce significant safety and misuse risks during multi-turn interactions with external environments. Existing safety mechanisms mainly rely on external guardrails, which have a limited ability to perform fine-grained behavioral control during execution. Meanwhile, recent mechanistic interpretability methods for LLM safety are mostly confined to single-turn or jailbreak-style QA settings, limiting their ability to capture the evolving risk dynamics of multi-turn agent execution. In this paper, we investigate the safety of multi-turn coding agents from an internal perspective. We propose AgentLens (Mechanistic Subspace Intervention and Steering), a white-box defense framework that performs runtime safety detection and representation-level mitigation for coding agents. Unlike conventional agent guardrails, AgentLens detect harmful execution states from step-level hidden representations and mitigate unsafe behavior by intervening in a 10-dimensional subspace within a single layer. To support this research, we introduce the Mechanistic Agent Safety (MAS) benchmark, comprising comprehensively annotated multi-turn execution trajectories across 194 tasks using LLaMA-3.1-8B, Qwen-2.5-7B, and Gemma-2-9B. Extensive experiments show that AgentLens achieves strong safety detection performance, provides preliminary evidence for lookahead risk anticipation, and substantially reduces harmful actions of the coding agent, establishing a foundation for applying mechanistic interpretability to dynamic LLM agent safety. The code is available at: https://github.com/EddyLuo1232/AgentLens --- ### Clipping the Price of Adaptivity at the Tail Authors: Itai Kreisler, Yair Carmon, Oliver Hinder Published: 2026-06-21 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2606.22669v1 Adaptive stochastic convex optimization (SCO) methods face a fundamental ``price of adaptivity'' barrier: under the standard set of assumptions, they cannot efficiently adapt to large uncertainty in both the initial distance to optimality and the Lipschitz constant. We circumvent this barrier by requiring a small amount of additional structure common to many learning problems. Specifically, we assume that the objective decomposes into a model and a loss function, enabling us to intervene by modifying the model's output before it passes to the loss function. Under this assumption, we design a method that clips the learned model output in tail events where it deviates too much from the output of a fixed reference model. Our method matches the optimal bounds for known-parameter SCO up to logarithmic factors in the uncertainty in the distance and Lipschitz parameters, thus efficiently adapting to large uncertainty in both. --- ### From Complaint Narratives to Monetary Relief: A Hybrid Machine Learning Framework for CFPB Consumer Complaints Authors: Zhuoer Wang, Sizhen Zhu, Xiongyu Chen Published: 2026-06-21 Categories: cs.CE, cs.LG arXiv: https://arxiv.org/abs/2606.22664v1 Consumer financial complaints provide a valuable source of information for identifying service failures, dispute frictions, and operational deficiencies in consumer-facing financial institutions. This paper proposes a hybrid machine learning framework for predicting monetary relief outcomes using Consumer Financial Protection Bureau complaint data. We formulate the task as an imbalanced binary classification problem, where complaints closed with monetary relief are treated as compensable outcomes. The proposed framework integrates multiple sources of predictive information, including complaint narrative text, LDA-based topic representations, interpretable text-engineered features, and structured categorical attributes such as company and state. An XGBoost classifier is trained using a temporal train-test split, with earlier complaints used for model development and more recent complaints reserved for out-of-sample evaluation. Compared with a TF-IDF baseline, the proposed framework substantially improves predictive performance, increasing AUC-ROC from 0.69 to 0.78 and improving PR-AUC under class imbalance. Feature importance analysis shows that textual signals, latent complaint topics, and company identity all contribute meaningful predictive information. In particular, company-level effects reveal systematic variation in complaint resolution patterns across financial institutions. These findings suggest that consumer complaint narratives can serve as alternative data for monitoring consumer harm, identifying firm-level operational weaknesses, and supporting early-stage risk surveillance in consumer finance. --- ### LSTM Variants for Chaotic Dynamical Systems: An Empirical Study on the Lorenz Attractor Authors: Ruslan Gokhman Published: 2026-06-21 Categories: cs.LG, eess.SY arXiv: https://arxiv.org/abs/2606.22662v1 Forecasting chaotic dynamical systems such as the Lorenz attractor is notoriously difficult: small numerical errors are amplified exponentially over long autoregressive rollouts. We study seven recurrent and convolutional architectures for the AI-DEEDS 2026 Chaotic Systems Challenge: a vanilla LSTM, an LSTM with additive attention, a Bidirectional LSTM (BiLSTM), a BiLSTM trained with the Huber loss, a Temporal Convolutional Network (TCN), a CNN front-end followed by an LSTM, and a CNN front-end followed by a BiLSTM. All models share the same pre-processing, sequence length, and rollout procedure, isolating the contribution of each design choice. The challenge scores predictions on a 0-100 scale where higher is better. We obtain leaderboard scores between 45.72 and 58.81, with the BiLSTM trained with Huber loss being the strongest configuration. Two findings stand out: (i) adding additive attention to the unidirectional baseline degraded performance by over ten points, and (ii) prepending a CNN front-end to either an LSTM or a BiLSTM did not help and slightly hurt the score. Per-pair RMSE measurements confirm that the BiLSTM family generalizes better in the harder pairs (6-7), while the LSTM + Attention model collapses there (RMSE up to 8.94 on pair 6). We discuss why bidirectional context and a robust loss help in chaotic regimes while attention and CNN front-ends fail in this setting. --- ### Prompting Diffusion Models for Zero-Shot Instance Segmentation Authors: Irem Zeynep Alagöz, Nils Morbitzer, Andrea Ramazzina, Nassir Navab, Federico Tombari, Stefano Gasperini Published: 2026-06-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.22660v1 Several disruptive research directions have recently emerged in computer vision, including foundation models achieving previously unseen zero-shot performance in scene understanding, even interactively, and generative models that synthesize extremely realistic images. The latter have also been shown to be highly effective in scene understanding tasks thanks to their rich priors. However, for promptable segmentation, foundation models struggle with accurately segmenting an object's region, leading to false positives and over-segmentation. Notably, early attempts that leverage generative priors use prompts only during post-processing, yielding suboptimal segments because the process is agnostic to the user input. In this paper, we target these limitations with Prompt2Seg, a spatial conditioning framework for diffusion-based segmentation. Prompt2Seg augments a frozen diffusion segmentation model with a conditioning branch. Our approach takes spatial prompts, represented as 2D Gaussians or confidence maps, as explicit input signals, training the model to respond directly to user intent. Fine-tuned on a deliberately constrained set of object categories drawn from Hypersim and Virtual KITTI 2, Prompt2Seg generalizes zero-shot to a wide range of unseen object types and visual domains. We evaluate on seven datasets ranging from standard benchmarks to more challenging domains, including paintings, egocentric views, and X-ray data. Furthermore, we demonstrate that Prompt2Seg consistently outperforms the underlying diffusion segmentation backbone across all benchmarks. Our results suggest that the rich priors encoded in generative pretraining, combined with principled spatial conditioning, offer a compelling path toward broadly generalizing interactive segmentation without large-scale mask supervision. --- ### Confidently Wrong: Severity-Aware Calibration of Prompt-Injection Detectors under Attack Shift Authors: Md Anas Biswas Published: 2026-06-21 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2606.22659v1 Prompt-injection detectors are deployed as guards: a model scores an input and a downstream system trusts or blocks it on that score. I study the confidence of these scores, not only their accuracy, when the attack distribution shifts away from the clean benchmark on which the operating point was chosen. I evaluate three released detectors, ProtectAI-v2 and two Prompt-Guard-2 checkpoints, at a single source-calibrated threshold that I freeze and transport across five shifts. I report a severity metric S, how confident a detector is on the attacks it misses, alongside the false-negative rate and discrimination. Across every shift and every detector, severity on the missed attacks stays between 0.99 and 1.00 while the false-negative rate ranges from 0.01 to 0.97: when these detectors miss, they miss with near-certainty. All three confidently pass indirect behavior-hijack injection, a blind spot unanimous across two vendors and a fourfold size range. Standard pooled calibration error does not register this; one detector it rates well-calibrated, at 0.06, is miscalibrated at 0.91 on the attacks alone. Run against live models, the missed injections leak the majority of working exploits, passing them at the rate they catch others. A controlled experiment traces the cause to content-keying rather than injection structure, an instruction-tuned model used as a judge shows the same hijack blind spot, and a black-box rewriter exploits the content-keying to manufacture working confident misses, most effectively on the most dangerous attack category. Code and data are public. --- ### Foundation Models for Epileptogenic Zone Identification in Drug-Resistant Epilepsy Authors: Thi Kieu Khanh Ho, Thomas Lai, Petr Klimes, Jan Cimbalnik, Martin Pail, Milan Brazdil, Birgit Frauscher, Narges Armanfard Published: 2026-06-21 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.22657v1 Accurate identification of the epileptogenic zone (EZ) is essential for seizure freedom after resective surgery in drug-resistant epilepsy, yet seizure freedom rates remain below 50%. We developed EpiiSLM, a dual foundation model system for EZ identification with stereo-electroencephalography (sEEG), by training a signal foundation model on 104,990 minutes of sEEG recordings from the Montreal Neurological Institute & Hospital, while leveraging all recordings regardless of surgical outcome and anchoring EZ biomarker extraction on non-epileptic signals. A language foundation model then integrates sEEG-derived outputs with multimodal clinical information to produce interpretable predictions. Under leave-one-patient-out evaluation, EpiiSLM achieved 0.978 contact-level positive predictive value (PPV), outperforming the seizure onset zone(SOZ)-as-EZ baseline by 15.1% (p < 0.05), and 100% region-level accuracy; on an external dataset, EpiiSLM achieved 0.857 contact-level PPV. EpiiSLM requires only one night of interictal sleep data, suggesting potential to reduce invasive sEEG monitoring duration and improve surgical outcomes. --- ### A Markov Chain Approach to Preference Alignment Authors: Takuya Koriyama, Tengyuan Liang Published: 2026-06-21 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2606.22652v1 We propose Markov Chain from Human Feedback (MCHF), an elementary approach for aligning generative models from pairwise human preferences. Unlike Reinforcement Learning from Human Feedback (RLHF), which reduces comparisons to a scalar reward, and Nash Learning from Human Feedback (NLHF), which preserves pairwise utilities through a KL-regularized minimax optimization, MCHF uses pairwise preferences directly to define a transition mechanism over model outputs. Given a pairwise utility $U(x,y)$, which quantifies human preference for $y$ over $x$, and a reference probability distribution $μ_{\mathsf{ref}}$, we define a Markov kernel $\mathsf{P}(x, dy)\propto \exp(U(x,y))μ_{\mathsf{ref}}(dy)$, and take the Markov chain starting from $μ_{\mathsf{ref}}$ as an iterative alignment procedure. We show that MCHF converges geometrically fast to the stationary distribution, with a convergence rate governed by the seminorm $\|U\|_\oplus=\inf_{g,f\in L^\infty(μ_{\mathsf{ref}})}\|U-g\oplus f\|_\infty$, which quantifies the non-transitive structure of the pairwise utility. We further show that a mirror-descent algorithm for NLHF satisfies an analogous structure-adaptive convergence guarantee. Finally, through a perturbation analysis, we prove that when $\|U\|_\oplus$ is small, MCHF and NLHF agree up to first order around an RLHF solution, which yields a unified view of reward-based, game-theoretic, and Markovian approaches to alignment. In particular, for two natural algorithms that converge to the MCHF/NLHF equilibria, we show that the first step of MCHF and NLHF recovers the RLHF solution based on the column-sum reward $\hat{f}(y)=\int μ_{\mathsf{ref}}(dx) U(x, y)$, and starting from the second iteration, both algorithms incorporate the same linear functional of the residual $U-(-\hat f)\oplus \hat f$, which captures the non-transitive structure of the pairwise utility $U$. --- ### MaRS: Robust Out-of-Distribution Detection via Mahalanobis Residual Scoring Authors: Francesco Di Salvo, Sebastian Doerrich, Christian Ledig Published: 2026-06-21 Categories: cs.CV, cs.LG, eess.IV arXiv: https://arxiv.org/abs/2606.22649v2 Foundation models provide highly descriptive representations for medical images, yet their reliability degrades under distribution shifts arising from changes in patients, devices, or acquisition conditions. Reliable out-of-distribution (OOD) detection is therefore essential for safe deployment. Recent post-hoc detectors efficiently exploit frozen embeddings (e.g., kNN), whereas reconstruction-based OOD detection in latent feature space has seen limited adoption due to inconsistent performance. In this work, we show that the limitation of reconstruction-based methods in latent space does not stem from poor reconstruction quality, but from how reconstruction errors are scored. Standard L2 residual norms collapse the anisotropic residual structure, thereby suppressing informative deviations. To address this limitation, we introduce MaRS (Mahalanobis Residual Scoring), a label-free OOD detector that learns an in-distribution manifold using a lightweight autoencoder and measures deviation via a Mahalanobis distance on reconstruction residuals, yielding variance-aware OOD scores. Across three imaging modalities, multiple types of distribution shift, and different model families and scales, MaRS outperforms established confidence-, distance-, and reconstruction-based baselines, while remaining fully post-hoc and lightweight. The code is available at https://github.com/francescodisalvo05/mars. --- ### RAVEN: Agentic RAG for Automated Vulnerability Repair Authors: Varun Gadey, Zijie Liu, Alexandra Dmitrienko Published: 2026-06-21 Categories: cs.CR, cs.LG, cs.SE arXiv: https://arxiv.org/abs/2606.22647v1 Automated vulnerability repair has emerged as a promising direction to mitigate the growing number of software vulnerabilities. Recent advances in Large Language Models (LLMs) have further accelerated research in automated repair. However, existing frameworks remain largely restricted to memory-related vulnerabilities and locally repairable vulnerability settings, leaving generalization to unseen vulnerability types underexplored. Their evaluations are often limited to a single programming language, and largely rely on proprietary models. In this paper, we propose RAVEN, a scalable, efficient and autonomous framework that integrates an agentic retrieval-augmented generation (RAG) pipeline with controlled iterative repair in a unified framework. The framework utilizes open-source LLMs in a fully locally deployable setting with limited GPU requirements, while building a multi-faceted retrieval pipeline to retrieve historically relevant vulnerability fixes and guide the patch generation. In addition, RAVEN introduces a dedicated Curator Agent that retrieves cross-file dependencies from the target repository, to fix complex vulnerabilities that cannot be addressed using local vulnerable code alone. We evaluate RAVEN on 160 real-world CVE vulnerabilities across diverse vulnerability types, two programming languages, unseen CWE categories, and out-of-distribution settings. RAVEN achieves an overall repair success rate of 83.13%, outperforming all existing state-of-the-art repair frameworks, while also demonstrating strong generalization capabilities and maintaining the repair cost negligible. --- ### Statistical Inference for Misspecified Contextual Bandits Authors: Yongyi Guo, Ziping Xu Published: 2026-06-21 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2606.22639v1 Contextual bandit algorithms have transformed modern experimentation by enabling real-time adaptation for personalized treatment. Yet these advantages create challenges for statistical inference due to adaptivity. We study inference with contextual-bandit data without assuming a well-specified outcome model. In this setting, we show a previously overlooked issue: standard algorithms such as LinUCB may fail to stabilize under misspecified working models, leading to non-Gaussian estimator behavior and invalid inference. This issue is practically important, as misspecified working models -- such as approximations of complex dynamical systems -- are often employed by online agents in real-world adaptive experiments to balance reward, computational tractability, and robustness. We develop an inverse-probability-weighted Z-estimation framework for a broad class of marginal moment targets, including projection parameters, structural parameters with noisy contexts, and off-policy values. We identify a stability condition tailored to this framework, scaled inverse-propensity convergence, under which the IPW-Z estimator is consistent and asymptotically normal with a consistent sandwich variance estimator. We further establish sufficient conditions for scaled inverse-propensity convergence for several policy classes, including multi-armed bandit algorithms and smooth contextual allocation policies. Simulations and a HeartSteps V1 real-data-calibrated application show reliable coverage and competitive performance across multiple targets. Overall, our results highlight the importance of stability-aware adaptive design for valid post-experiment inference. --- ### Design and Development of a Neuromorphic Silicon Suite: PVT Sensing, Stochastic LIF Inference, On-Chip STDP Learning, and Crossbar Programming Authors: Poornima Kumaresan, Santhosh Sivasubramani Published: 2026-06-21 Categories: cs.ET, cs.AR, cs.NE arXiv: https://arxiv.org/abs/2606.22635v1 Edge neuromorphic systems need compact, configurable hardware that combines probabilistic inference, local learning, and an interface to emerging analogue memory. We present four interface-compatible digital IP blocks implemented as standard-cell CMOS on the SkyWater 130 nm process: a process, voltage and temperature (PVT) sensor built from five selectable ring oscillators that also provides a jitter-based true-random-number generator and a frequency-bounds health monitor; a stochastic leaky integrate-and-fire (LIF) neuron with a configurable LFSR, a programmable activation table, and a refractory period; an on-chip spike-timing-dependent plasticity (STDP) controller with a programmable curve and reward-modulated, eligibility-trace, and anti-Hebbian modes; and a memristive-crossbar controller supporting forming, set, reset, read, and automated current-voltage sweep with current-compliance limiting and half-select biasing. All four blocks share a common serial peripheral interface (SPI) register file; the sensor also exposes a parallel readout. Each occupies a single tile at a 50 MHz target. The suite was verified with 99 cocotb tests at register-transfer and gate level (all passing) and taken through an open standard-cell flow, then submitted for tapeout via the Tiny Tapeout shared-silicon programme. Mapped to the open cell library, each block occupies a post-synthesis cell area of 9.3 to 10.6 thousand square micrometres, places at 61 to 70 per cent tile utilisation, meets the 50 MHz constraint with positive setup and hold margin after clock-tree synthesis, and draws an estimated 0.64 to 0.70 mW under a default switching-activity assumption. The contribution is a coherent, openly released set of building blocks unified by one register interface and one verification flow. All results are from simulation and the implementation flow; no fabricated silicon is reported. --- ### Learning Entropy Signature for Image Representation and Classification Authors: Jan Glaser, Ivo Bukovsky, Noriyasu Homma, Marcel Jirina Published: 2026-06-21 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2606.22634v1 Learning Entropy (LE) has recently been extended to image analysis through Spatial Learning Entropy Maps (SLEMs), which are two-dimensional LE distributions that highlight unusually high learning activity across an image. Unlike conventional image descriptors, SLEMs are generated by incremental, sample-wise learning of a pretrained feedforward MLP network, where local pixel neighborhoods are presented sequentially in a fixed spatial order to predict the corresponding central pixels. Consequently, the learning activity at each image location depends not only on its local structure but also on the knowledge acquired from previously processed locations. This paper introduces Learning Entropy Signatures (LES), an image descriptor derived from SLEM using the K largest LE locations. LES captures the spatial organization of learning-relevant image structures and provides a compact representation of image content based on learning weight behavior. Experimental evaluation on image classification tasks shows that a relatively small number of K largest LE locations preserve substantial discriminative information. The results indicate a close relationship between the learning of neural weights and information relevance, extending the role of Learning Entropy from time series to images and, within images, from structural point extraction to compact image representation and classification. --- ### Confident but Conflicted: Internal Uncertainty and Cognitive Dissonance Resolution in LLMs Authors: Weihong Qi, Kristina Lerman Published: 2026-06-21 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.22633v1 Large language models (LLMs) frequently encounter inputs that disagree with their prior outputs, through user pushback, retrieved documents, or web search results. While the way they resolve such conflicts -- a process we frame as cognitive dissonance resolution -- has been characterized behaviorally, its connection to internal model uncertainty is not well understood. To study this systematically, we vary persuasion attempts along two dimensions, source authority and evidence quality, across 12 health-science claims of stratified epistemic status. Dissonance can be resolved through persuasion, backfire, or immunity. We introduce Trust Elasticity (TE), an econometrics-inspired measure of how readily a model is persuaded toward conflicting evidence. Across four LLMs, TE varies substantially, while clearly false claims elicit near-zero TE across all models. On two open-weight models, we further find that this variation is associated with two complementary internal uncertainty indicators, Confidence Miscalibration in Qwen and Internal Uncertainty Change in Llama. These results link cross-model behavioral variation to a measurable internal property and point to interventions targeting internal uncertainty as future work. --- ### 4DVLT: Dynamic Scene Understanding with Worldline-Centered Vision-Language Tracking Authors: Chaoyue Li, Boxue Yang, Shengyao Zhou, Haoyang Wu, Rui Qian, Linfeng Zhang Published: 2026-06-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.22631v1 4D dynamic scene understanding requires grounding language to a persistent worldline that binds identity, metric 3D motion, and synchronized multi-view 2D projections. Existing paradigms capture only part of this structure: large multimodal models reason over rich visual evidence but rarely preserve metric topology, while vision-language tracking remains tied to fragmented 2D or 3D outputs and local continuation. We therefore introduce \textbf{4DVLT}, a worldline-centered task for instruction-conditioned 4D dynamic scene understanding in fully observed multi-view video, and \textbf{Instruct-4D}, a benchmark with 129.4K question-answer pairs, 64.7K target entities, 851 scenes, and 9 reasoning-oriented query types. To address this setting, we present \textbf{4DTrack}, which casts instruction-conditioned tracking as graph-conditioned worldline inference through an object-centric 4D state graph, metric-guided routing, bidirectional decoding, and kinematic calibration. On Instruct-4D, 4DTrack-Qwen3.5-9B reaches 62.68 $\mathrm{TGA}_{\mathrm{Top1}}$ and surpasses the best adapted VLT baseline by 19.62 points. These results show that worldline-centered modeling improves both target grounding and recovered worldline quality. The project page is available at https://github.com/mikubaka88/4DVLT. --- ### Scalable Maximum Entropy Reinforcement Learning for Diffusion Policies via Adjoint Matching Authors: Serge Thilges, Onur Celik, Denis Blessing, Emiliyan Gospodinov, Gerhard Neumann Published: 2026-06-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.22630v1 Diffusion policies have recently emerged as a powerful paradigm for representing complex action distributions in reinforcement learning (RL). However, their application to online RL remains limited by the challenge of scalable training in the absence of ground-truth data, where standard optimization techniques such as score matching are not directly applicable. In this work, we introduce a highly efficient algorithm for optimizing diffusion policies by leveraging recent advances in stochastic optimal control. Our approach is based on adjoint matching, which enables simulation-free training and circumvents the need for explicit likelihood estimation or costly backpropagation through the diffusion process. Furthermore, we propose several extensions that improve the robustness and stability of the method in practical settings. Empirical results demonstrate that our approach achieves competitive performance while significantly reducing computational overhead, making diffusion policies more viable for online RL scenarios. --- ### Orthogonal Representation Editing: Decoupling Semantic Entanglement in Batch Knowledge Editing of LLMs Authors: Wenhao Yu, Zhicong Lu, Bo Lv, Fangyin Ma, Kaiwen Wei, Shihao Yang, Nayu Liu Published: 2026-06-21 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.22627v1 Knowledge editing aims to efficiently update factual information in Large Language Models (LLMs) without full retraining. However, existing methods still suffer from performance degradation in batch knowledge editing. We identify that semantic representation entanglement, such as overlapping concepts and shared syntactic patterns, accumulates interference in the representation space and reduces editing precision. To bridge this gap, in this paper, we propose Orthogonal Representation Editing (ORE), which performs edits in the hidden representation space of LLMs by constructing a general semantic subspace and enforcing orthogonal constraints on edit vectors, effectively decoupling semantic entanglement. Furthermore, we introduce a gated non-linear representation head to enable adaptive learning of editing locations and precise control over knowledge injection. Extensive experiments show that ORE outperforms existing methods and achieves superior performance in cross-lingual knowledge editing scenarios. We release our code at https://github.com/YVVH/ORE. --- ### DR-Mamba: Automatic Inference-Time Domain Adaptation for Document Image Binarization via Sample-Conditioned Detail-Background Suppression Authors: Sheng-Wei Chan, Jen-Shiun Chiang Published: 2026-06-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.22625v1 Degraded document image binarization is sensitive to domain shifts caused by paper aging, bleed-through, stains, shadows, and uneven illumination, and the foreground-background separation of recent learning-based methods can become unstable on unseen degradation domains. We propose DR-Mamba, a sample-conditioned detail-background suppression framework that performs automatic inference-time domain adaptation for document image binarization. Unlike test-time adaptation methods that require gradient updates or auxiliary data at inference, DR-Mamba adapts to each input document through input-dependent gates within a single forward pass, requiring no target-domain labels, no fine-tuning, and no test-time parameter updates. Instead of using Mamba-style selective scanning as a single generic feature path, DR-Mamba reinterprets it as fast-slow route modeling: a fast detail route captures local stroke structures, while a slow background route accumulates spatially persistent degradation responses. The two routes are integrated through an input-dependent subtractive gate that explicitly suppresses background interference rather than fusing features by addition or concatenation. We further add full-resolution detail-guided reconstruction and thin-stroke-aware supervision to recover fine strokes lost during downsampling. Evaluated under a leave-one-year-out protocol on DIBCO-style benchmarks, where each held-out year is treated as an unseen degradation domain, DR-Mamba shows that per-document, per-location subtractive suppression improves cross-domain robustness, with particularly strong performance on the most severely degraded held-out fold. --- ### Multi-Level Resistive Synapses for On-Chip Neural Networks: A Physics-Based Design of a Memristive Crossbar Fabric with Quasi-Continuous Conductance States Authors: David Alejandro Trejo Pizzo Published: 2026-06-21 Categories: cs.AR, cond-mat.mes-hall, cs.ET, cs.NE, physics.app-ph arXiv: https://arxiv.org/abs/2606.22621v1 Building on resistive communication, this paper presents a physics-based design of an on-chip neural network with multi-level memristive synapses supporting a dense spectrum of conductance states. Derived from ionic transport physics, we develop a state-variable model and quantify storable sub-levels under thermal noise, drift, and quantized conductance. We assemble these devices into a 1T1R crossbar fabric, derive the linear algebra of analog vector-matrix multiplication (VMM) under wire resistance, and design a differential synapse for signed weights. A multilayer pipeline executes inference, backpropagation, and weight updates physically in the analog domain. We derive the in-situ outer-product learning rule, its discretization onto the conductance lattice, and the resulting quantization noise. We provide energy, area, capacity, and inter-tile models, showing this substrate is ideally suited for large language models (LLMs). Our design eliminates weight movement, surpassing binary ReRAM and traditional CMOS. We detail the material stack (HfO_2-based), the FEOL/BEOL CMOS foundry-integration flow, a self-contained SPICE model, the complete memristive-FPGA neuromorphic system, and an in-memory self-attention engine with current-mode translinear softmax. Finally, a ternary BitNet datapath shows projected per-token efficiency orders of magnitude better than advanced CPUs or GPUs. The result is a self-contained hardware-native blueprint for a high-density, analog, in-memory neural processor. --- ### Federated Learning for Global Carbon Emission Forecasting: A Hybrid Time-Series Approach with Statistical and Neural Models Authors: Attia Qammar, Qazi Haseeb Yousaf, Ali Azam, Ammar Ahmed, Abdenacer Naouri, Tianrui Li Published: 2026-06-21 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.22618v1 Climate change, primarily driven by carbon dioxide (CO2) emissions, requires accurate forecasting tools to support effective mitigation policies and sustainable development strategies. Existing forecasting approaches typically rely on centralized data collection, which is often restricted by privacy regulations and the distributed nature of emission data across countries and industrial sectors. This paper proposes a novel federated hybrid forecasting framework that integrates ARIMA-based trend modeling, GARCH-based volatility modeling, LSTM-Attention temporal representation learning, and XGBoost prediction within a privacy-preserving federated learning environment. The proposed framework enables collaborative learning among distributed clients without requiring the exchange of raw data. Experimental evaluation across 14 clients demonstrates strong forecasting performance, achieving client R2 values between 0.50 and 0.97 with an average of 0.73, RMSE values ranging from 0.06 to 2.35 with an average of 1.21, and MAPE values between 1.5% and 11.3% with an average of 6.5%. The results indicate that the proposed framework provides an accurate, scalable, and regulation-compliant solution for collaborative carbon-emission forecasting. --- ### OmniSpace: Efficient Geometry Awareness for Autonomous Vehicles MLLMs Authors: Hao Vo, Phu Loc Nguyen, Khoa Vo, Sieu Tran, Duc Minh Nguyen, Ngo Xuan Cuong, Nghi D. Q. Bui, Anh Nguyen, Duy Minh Ho Nguyen, Ngan Le Published: 2026-06-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.22617v2 Multimodal Large Language Models (MLLMs) have achieved remarkable performance on 2D visual tasks, yet enhancing their spatial intelligence for real-world applications such as Autonomous Vehicles (AV) remains an open challenge. Existing geometry-aware MLLMs typically rely on auxiliary 3D models at inference time, introducing pipeline complexity and the risk of cascading failures. In this paper, we present OmniSpace, a simple yet effective plug-and-play paradigm for geometry-aware spatial reasoning from purely 2D observations. Motivated by our finding that current MLLMs are bottlenecked by weak cross-view correspondence and depth estimation, OmniSpace introduces a Camera Pose Injector, a Multi-view Epipolar Attention module, and a 3D Geometric Distillation objective that jointly address these two limitations by transferring geometric knowledge into the model. Extensive experiments show that OmniSpace surpasses existing methods on planning benchmarks (nuScenes, Bench2Drive), risk detection (nuInstruct), language (Omnidrive), and generalization (DriveBench). --- ### OmniSpace: Efficient Geometry Awareness for Autonomous Vehicles MLLMs Authors: Hao Vo, Phu Loc Nguyen, Khoa Vo, Sieu Tran, Duc Minh Nguyen, Ngo Xuan Cuong, Nghi D. Q. Bui, Anh Nguyen, Duy Minh Ho Nguyen, Ngan Le Published: 2026-06-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.22617v1 Multimodal Large Language Models (MLLMs) have achieved remarkable performance on 2D visual tasks, yet enhancing their spatial intelligence for real-world applications such as Autonomous Vehicles (AV) remains an open challenge. Existing geometry-aware MLLMs typically rely on auxiliary 3D models at inference time, introducing pipeline complexity and the risk of cascading failures. In this paper, we present OmniSpace, a simple yet effective plug-and-play paradigm for geometry-aware spatial reasoning from purely 2D observations. Motivated by our finding that current MLLMs are bottlenecked by weak cross-view correspondence and depth estimation, OmniSpace introduces a Camera Pose Injector, a Multi-view Epipolar Attention module, and a 3D Geometric Distillation objective that jointly address these two limitations by transferring geometric knowledge into the model. Extensive experiments show that OmniSpace surpasses existing methods on planning benchmarks (nuScenes, Bench2Drive), risk detection (nuInstruct), language (Omnidrive), and generalization (DriveBench). --- ### SkillAudit: From Fixed-Suite Benchmarking to Skill-Centered Assessment Authors: Dexu Yu, Youhua Li, Zhaoyang Guan, Xianhao Lin, Jining Luan, Zihao Rao, Xuanqi Lan, Yang Ran, Bo Lan, Nai-Xin Zhai, Hanwen Du, Junchen Fu et al. Published: 2026-06-21 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.22613v1 Agent skills have become a practical way to extend large language model agents, but the growing skill ecosystem still lacks a reliable way to judge whether a skill is worth deploying. Existing evaluation methods remain largely anchored to fixed task suites, assessing skills through performance on predefined tasks and environments. As skill marketplaces expand, this paradigm becomes inadequate: fixed suites can conflate a skill's marginal contribution with backbone strength and miss its value when tasks fall outside the skill's intended scope. We introduce SkillAudit, an end-to-end framework for skill-centered assessment that takes an arbitrary agent skill as input and automatically generates a comprehensive, multi-dimensional evaluation report spanning utility, efficiency/cost, and safety. SkillAudit focuses on the skill artifact itself and constructs capability-aligned evaluation tasks directly from the skill package. The generated tasks are conducted in isolated sandbox environments to collect execution evidence, followed by automated checks with LLM-based judging to produce auditable results. To dissect the agent skills, we propose the baseline comparison principle to measure utility and efficiency/cost, and introduce a two-stage detection paradigm combining static semantic analysis with dynamic runtime verification to assess safety risks. After scanning top-ranked real-world skill packages spanning 23 occupational categories, we found that over 7% of skills are at risky status. --- ### PaperClaw: Harnessing Agents for Autonomous Research and Human-in-the-Loop Refinement Authors: Weiwei Ye, Hangchen Liu, Dongyuan Li, Renhe Jiang Published: 2026-06-21 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.22610v1 Large language models have become capable reasoners and tool users that write and run code and search the literature, which makes automating the research process itself a realistic goal. We present PAPERCLAW, a harnessed multi-agent system that carries a project autonomously, from a field of study to a finished paper. PAPERCLAW curates a domain from a field's live literature, datasets, and code; brainstorms it into an idea with a pre-registered main-result contract; and drives a stoppable hypothesis map through an iterative propose, test, reflect loop that grows only from measured verdicts and halts once the evidence supports the idea, at which point it writes a venue-compliant paper. A full-lifecycle memory keeps each stage in a single living record, so a long run can be paused, inspected, and resumed without losing context. At the centre is an in-cycle research assistant with research tools and skills: it can drive the whole pipeline on its own, while the same interface lets a person step in at any stage, turning a first autonomous draft into a stronger paper through human-in-the-loop refinement. Throughout, PAPERCLAW keeps its output grounded and checkable, citing only references validated against open scholarly indexes and reporting results that genuinely ran. An evaluation with an LLM judge finds that PAPERCLAW produces strong papers both fully autonomously and with human-in-the-loop refinement. --- ### Automated sign detection across the Electronic Babylonian Library: A large-scale dataset and end-to-end cuneiform OCR pipeline Authors: Wentao Che, Esteban Garcés Arias, Asim Niaz, Andreas Bender, Enrique Jiménez Published: 2026-06-21 Categories: cs.CV, cs.CL arXiv: https://arxiv.org/abs/2606.22608v1 Learning to read cuneiform tablets is an extremely demanding task; consequently, of the roughly half million excavated tablets, only a small fraction has been analysed by Assyriologists. Computer vision offers a promising avenue for decipherment but requires large, densely annotated datasets. To address this limitation, the largest annotated cuneiform sign dataset to date is used, and a Deformable Detection Transformer (DETR)-based object detection model is evaluated under two class granularities of 173 and 106 classes. The proposed system integrates automatic tablet-side extraction, heuristic line grouping, and n-gram-based textual similarity evaluation to bridge visual sign detection and textual structure, and achieves consistent improvements of up to 28-37% over prior work on COCO-style detection metrics. At inference, the method is applied to 87,668 tablet fragments from the Electronic Babylonian Library (eBL) corpus, producing nearly 2.9 million sign detections. Although the approach operates without linguistic priors and remains sensitive to tablet damage and layout variability, it provides a scalable and interpretable foundation for corpus-wide cuneiform analysis and supports future integration with multimodal and linguistic modelling frameworks. --- ### Sub-Billion, Super-Frontier: Small Language Models Rival Zero-Shot Frontier LLMs on General and Literary Relation Extraction Authors: Despina Christou, Grigorios Tsoumakas Published: 2026-06-21 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.22606v1 Large language models (LLMs) achieve strong relation extraction (RE), but their computational demands and reliance on proprietary APIs limit deployment in resource-constrained or privacy-sensitive settings. We investigate how far small language models (SLMs) can close this gap across general-domain and literary text. We evaluate five models from 360M to 3B parameters under three domain-composition regimes and two prompt-conditioned tuning styles (30 configurations), comparing them with zero-shot frontier LLMs and a discriminative RoBERTa baseline. Across nine benchmarks, the best sub-billion model, Qwen2.5-0.5B fine-tuned on pooled general-domain data, achieves a general-domain positive-class micro-F1 of 0.83, versus 0.69 for GPT-5.4 and 0.66 for Claude Sonnet 4.6 evaluated zero-shot. This does not imply that SLMs are intrinsically stronger; rather, targeted task adaptation enables 4-bit models deployable on a single consumer GPU to outperform general-purpose frontier systems under this protocol. An in-domain RoBERTa baseline also exceeds both frontier models, indicating that the gain stems from task adaptation rather than generative decoding. On literary RE, tuned SLMs reach 0.92 on the human-annotated Biographical benchmark versus 0.83 for GPT-5.4, and 0.833 versus 0.578 on the two-benchmark literary average. A targeted domain-adaptive pretraining case study yields no practically meaningful gain over supervised fine-tuning, while the cleanest within-family scale comparison shows only marginal improvement. These results show that, when task-specific data are available, compact task-adapted models can provide accurate, private, and hardware-efficient RE. --- ### A Theory-grounded Hybrid Neural Network Integrating Complementary Estimation Mechanisms for Stable Visual Object TrackingA Authors: Yancheng Zhou, Hanle Zheng, Lei Deng, Yujie Wu Published: 2026-06-21 Categories: cs.NE arXiv: https://arxiv.org/abs/2606.22604v1 Hybrid neural networks (HNNs) that integrate artificial neural networks (ANNs) with brain-inspired neural networks have achieved broad success across perception and control tasks. However, much of the current success is confined to neuron-scale hybridization, where discrete, spike-based coding fundamentally limits applicability to continuous-state estimation tasks. In neuroscience, continuous attractor neural networks (CANNs) represent continuous states through neural ensembles, pointing to a population-scale route for HNNs to address this limitation. Yet, principled methodologies for ANN-CANN integration remain largely underexplored. In this work, we propose a theory-grounded ANN-CANN hybridization framework and instantiate it as a hybrid tracking neural network (HTNN) for visual object tracking, a representative continuous-state estimation task. The framework aligns ANN response maps with CANN dynamics in the same state space, enabling the two heterogeneous branches to interact through the shared state representation. Furthermore, we uncover a functional bias-variance complementarity: data-driven ANNs provide asymptotically unbiased estimates, while CANN estimates are low-variance but temporally lagged. By operationalizing this complementarity, HTNN achieves stable and accurate tracking across nine visual tracking benchmarks, consistently outperforming single-network baselines and existing hybrid models. Notably, these performance gains are robustly maintained even under diverse environmental variations, including occlusion, motion blur, and background interference. Through this proof-of-concept study, our framework offers a generalizable foundation for advancing HNNs toward population-scale hybridization. --- ### Scalable Bayesian Additive Models for Stellar Flare Detection via Amortized Gaussian Process Inference and Hidden Markov Models Authors: Rodrigo Herrera, Vianey Leos-Barajas, Gwendolyn Eadie, Elizaveta Semenova, James Davenport Published: 2026-06-21 Categories: stat.ML, cs.LG, stat.AP, stat.CO arXiv: https://arxiv.org/abs/2606.22601v1 Gaussian Processes (GPs) are a powerful tool for Bayesian time-series modeling, yet their cubic computational cost remains a severe barrier for application to long, high-cadence datasets in astronomy. While specialized scalable solvers like Celerite elegantly reduce this scaling to linear time, repeatedly evaluating the exact likelihood during iterative Bayesian sampling is a bottleneck for developing more complex models, like hierarchical or additive models in which Celerite is only one component. To make this inference computationally tractable, we introduce a generative surrogate framework. By utilizing a Variational Autoencoder (VAE) to learn a compressed representation of the Celerite prior, we map highly correlated stochastic dependencies into a low-dimensional, isotropic manifold. This transition completely bypasses exact covariance operations, shifting the computational burden to a rapid neural network forward pass. Through an extensive simulation study, we show that the generative surrogate accurately reproduces the structural fidelity of exact physical kernels like Celerite. Finally, we demonstrate embedding our VAE approximation into an additive model that combines Celerite and a hidden Markov model (HMM) for stellar flare detection in time series data of stars. We evaluate the joint VAE+HMM architecture against the exact Celerite+HMM framework on empirical astrophysical time series and demonstrate that the proposed methodology achieves significant reductions in computational time, enabling the rigorous, large-scale characterization of stellar flares across massive data archives. --- ### On the Position Bias of On-Policy Distillation Authors: Yan Xie, Sijie Zhu, Tiansheng Wen, Bo Chen, Yifei Wang Published: 2026-06-21 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.22600v3 On-Policy Distillation (OPD) improves the learning efficiency of standard reinforcement learning through dense, token-level supervision from teachers. In the standard KL objective of OPD, token-level losses are uniformly averaged, implying equal weights for all tokens. However, we discover that not all tokens are created equal: as student rollouts grow longer, they deviate further from the teacher's distribution, leading to degraded supervision quality at later positions. As a result, OPD using only the first 30% of tokens can perform comparably to using all tokens, whereas OPD using only the last 30% of tokens barely learns anything. In this work, we provide a principled understanding of this issue through the lens of constrained optimization. Based on these insights, we derive Importance-Weighted On-Policy Distillation (IW-OPD), in which the weight assigned to each token depends on the accumulated discrepancy between the student's and teacher's distributions, naturally upweighting earlier tokens and downweighting later ones with larger deviations. We show that IW-OPD converges significantly faster than OPD, with better learning efficiency, and achieves better final performance than standard OPD in both same-size and cross-scale settings, improving performance up to 6.9 points on AIME-2025. --- ### MapReason-OSM: Can Vision-Language Models Make Graph-Verifiable Mobility Decisions from Street Maps ? Authors: Srinivas Venkatanarayanan, Clement Pakkam Isaac Published: 2026-06-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.22597v2 Vision-language models (VLMs) are increasingly used to read maps for logistics, delivery, and accessible navigation, where the output is an actionable decision (a route, a pin, a parking choice) that must respect the road network. Yet most map benchmarks grade free text or multiple-choice answers that cannot be verified against the underlying graph. We present MapReason-OSM, a benchmark and evaluation harness for graph-verifiable mobility decisions on self-rendered OpenStreetMap panels. We render fixed-style maps for ten U.S. downtowns at two aligned zoom scales, overlay a consistent marker grammar, and pair each panel with a hidden street graph and exact oracles, yielding 6,000 instances (12,000 panels across the two zooms) over 12 routing, facility-location, and visual disambiguation tasks. Models return structured decisions that we snap back to the graph and score for validity, legality, optimality, and constraint satisfaction, plus cross-zoom consistency. Across seven VLMs, models read maps and route simply but fail at graph cost reasoning (single-facility pin placement is near chance even for frontier reasoning models), and are frequently scale-inconsistent. We release the benchmark, harness, and deterministic generator. Code and data: https://github.com/Vi-Sri/mapreason-osm --- ### Training-free Task Classification for Multi-Task Model Merging Authors: Jungyong Son, Jinwook Jung, Sungyong Baik Published: 2026-06-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.22589v2 Ever since the advent of foundation models and the pre-training-finetuning paradigm, there have been numerous efforts to merge multiple task-specific experts into a single multi-task model. Prior work largely focuses on finding a single merged model, but it often underperforms individual experts due to parameter interference. To resolve this, dynamic model merging employs routing to activate task-relevant parameters per input. However, existing routers typically require either additional training with abundant labeled datasets or assume the access to task IDs of each input at inference time. In this work, we aim to close the gap to expert performance without additional training or task-ID-access assumption. To this end, we formulate routing as training-free task classification for each test input. Using singular value decomposition (SVD)-based low-rank manifold approximations for each task, SiM scores tasks by the projection residual of the test input feature onto each task manifold and routes accordingly. The task manifolds are pre-computable offline from a pretrained backbone using a small per-task support set (e.g., 32 examples per task) prior to merging process, requiring no router training and no data during the merging process. Moreover, SiM integrates seamlessly with subspace-/mask-based merging that represents task-expert via lightweight compressed task vectors, avoiding the need to store full expert parameters. Experiments across computer vision and natural language processing benchmarks under task-unknown inference demonstrate that SiM substantially improves merged-model performance and consistently narrows the gap to individual task experts. Our code is available at https://github.com/BAIKLAB/SiM --- ### Training-free Task Classification for Multi-Task Model Merging Authors: Jungyong Son, Jinwook Jung, Sungyong Baik Published: 2026-06-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.22589v1 Ever since the advent of foundation models and the pre-training-finetuning paradigm, there have been numerous efforts to merge multiple task-specific experts into a single multi-task model. Prior work largely focuses on finding a single merged model, but it often underperforms individual experts due to parameter interference. To resolve this, dynamic model merging employs routing to activate task-relevant parameters per input. However, existing routers typically require either additional training with abundant labeled datasets or assume the access to task IDs of each input at inference time. In this work, we aim to close the gap to expert performance without additional training or task-ID-access assumption. To this end, we formulate routing as training-free task classification for each test input. Using singular value decomposition (SVD)-based low-rank manifold approximations for each task, SiM scores tasks by the projection residual of the test input feature onto each task manifold and routes accordingly. The task manifolds are pre-computable offline from a pretrained backbone using a small per-task support set (e.g., 32 examples per task) prior to merging process, requiring no router training and no data during the merging process. Moreover, SiM integrates seamlessly with subspace-/mask-based merging that represents task-expert via lightweight compressed task vectors, avoiding the need to store full expert parameters. Experiments across computer vision and natural language processing benchmarks under task-unknown inference demonstrate that SiM substantially improves merged-model performance and consistently narrows the gap to individual task experts. --- ### Text2DSL: LLM-Based Code Generation for Domain-Specific Languages Authors: Alexander V. Kozachok, Alexander M. Nazimov, Shamil G. Magomedov Published: 2026-06-21 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2606.22586v1 Domain-specific languages (DSLs) are widely used for managing operating system security policies, yet manually authoring rules in such languages demands high expertise and is error-prone. This paper formalises the task of automatic DSL code generation from natural language descriptions - Text2DSL - as a distinct problem class, separate from Text-to-SQL and general-purpose code generation. We introduce the PolkitBench dataset comprising 4,204 verified natural-language-to-Polkit-rule pairs, each validated through a three-level AST-based pipeline. Controlled prompt experiments on two MoE models of different scale and provenance - GigaChat-10B-A1.8B (1.8B active parameters) and Nemotron-3-Nano-30B-A3B (3B active) - demonstrate the critical role of structured context (BNF grammar, API specification, permitted identifier vocabulary) for LLM-based DSL code generation. Across both models, supplying context raises syntactic validity to 98.6-99.4%, structural validity by +9.7 to +35.5 pp, and the CodeBLEU score by +60% to +95%. The consistency of the effect across models of different scale and provenance indicates that, for the Text2DSL class of problems, injecting a formal target-language specification into the prompt context is a robust enabling factor for high-quality generation without model fine-tuning. --- ### From CVE to CWE: Syscall-Based HIDS Generalisation Authors: Alexander V. Kozachok, Stanislav G. Vyugov, Shamil G. Magomedov Published: 2026-06-21 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2606.22581v1 Host intrusion detection systems (HIDS) based on system-call traces are typically trained and evaluated against individual Common Vulnerabilities and Exposures (CVE) instances. In operational settings, however, defenders need to recognise new exploits of an already known type of weakness. We empirically examine whether a one-class anomaly detector trained on the normal behaviour of a set of CVEs that share a Common Weakness Enumeration (CWE) class generalises to a different, unseen CVE inside the same class. Using six scenarios drawn from LID-DS-2021 and grouped into three CWE families (CWE-307 broken authentication, CWE-89 SQL injection, CWE-434 unrestricted file upload), we extract a 66-dimensional Peng-Guo-style feature vector per sliding window and train Isolation Forest and SGD One-Class SVM detectors with normal-only thresholds calibrated to fixed target false positive rates. We define and answer four research questions covering self-detection, asymmetric cross-CVE transfer, the value of a combined CWE-level normal profile, and the effect of feature filtering on transferability. The combined CWE-307 detector reaches F1 = 0.6976 at calibration target FPR = 0.05 (precision = 0.8994, recall = 0.5698), whereas CWE-89 and CWE-434 collapse to F1 <= 0.21 under the same protocol. Cross-CVE transfer turns out to be strongly direction-dependent and dominated by the breadth of the source normal profile rather than by the CWE label. We conclude that CWE-level generalisation in HIDS is empirically attainable for some but not all weakness families with current syscall features, and we argue that calibrated FPR is a methodological prerequisite for honest reporting in this setting. --- ### Stationary Robust Mean-Field Games under Model Mismatches Authors: Yue Wang Published: 2026-06-21 Categories: cs.LG, cs.GT arXiv: https://arxiv.org/abs/2606.22579v1 Deploying multi-agent reinforcement learning (MARL) in the real world is often limited by model mismatches between the training simulators and the true environment, which could be further amplified through strategic interactions and result in severe performance degradation upon deployment. Distributional robustness offers a principled response by optimizing policies against worst-case transition models drawn from an uncertainty set, but standard robust MARL frameworks become increasingly intractable as the number of agents grows. This paper develops an infinite-horizon, stationary mean-field game framework that incorporates distributional model uncertainty directly into the population-coupled dynamics. We establish a robust dynamic programming principle with a contractive Bellman operator and prove the existence of a stationary robust mean-field equilibrium via a fixed-point argument. We further develop the first concrete algorithm with convergence guarantees. We then connect the mean-field solution to a finite-population robust game whose ambiguity sets depend on the empirical distribution, showing that the mean-field equilibrium policy induces approximate equilibrium behavior as the population size increases. Under a contractive robust-dynamics regime, we further obtain explicit non-asymptotic error bounds. Numerical experiments further illustrate the qualitative and quantitative impact of robustness under multiple uncertainty models, validating our theoretical findings. --- ### Context-Aware Distillation and Ablation for Text2DSL Authors: Alexander V. Kozachok, Alexander M. Nazimov, Shamil G. Magomedov Published: 2026-06-21 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.22578v1 We extend our prior work on Text2DSL automatic generation of domain-specific language (DSL) code from natural language descriptions along two complementary axes. First, we replace prompt-only synthetic generation with context-aware distillation, in which a teacher large language model (DeepSeek-V4-Flash) operates under an explicitly defined structured context comprising a BNF grammar, an API specification, and a closed identifier vocabulary; the resulting corpus is verified by a two-tier pipeline combining AST validation through esprima and runtime acceptance through the production polkitd daemon and the pkcheck client. This scales the verified PolkitBench corpus from 4,204 to 10,073 natural-language-to-Polkit-rule pairs at 100.0% AST validity and 99.7% runtime pass rate. Second, we conduct the per-component factorial ablation of structured context that was identified as future work in the precursor study: eight conditions C0-C7 are evaluated on GigaChat-10B-A1.8B with the new corpus. Three findings emerge. (i) The new harder corpus collapses the baseline mode (Syntax Valid 97.6% -> 58.5%, Combined Score 0.482 -> 0.252), whereas the context-enhanced mode degrades only marginally (Syntax 98.6% -> 97.4%, Combined 0.801 -> 0.750), confirming that structured context is not a cosmetic improvement but a load-bearing mechanism. (ii) The best absolute condition is the full context C7 across all metrics, while the strongest partial conditions (C5 = BNF + Vocabulary, C6 = API + Vocabulary) both contain the vocabulary. (iii) A Shapley-style decomposition assigns the largest semantic-quality effect to the vocabulary (Combined +0.198), the largest structural-validity effects to API (+24.7 pp) and BNF (+22.3 pp). --- ### The Power of Light: Improving Synthetic-to-Real Domain Adaptation through Physically-Based Indirect Illumination Authors: Hooman Tavakoli Ghinani, Tatjana Legler, Martin Ruskowski Published: 2026-06-21 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.22574v2 While synthetic data generation resolves the manual labeling bottleneck in computer vision, minimizing the syn-to-real domain gap requires optimizing rendering variables. This paper presents a systematic study analyzing the impact of lighting configurations and background complexity on object detection performance. We introduce SmartSDG, an automated, reproducible pipeline built on NVIDIA Isaac Sim using Physically-Based Shading (PBS), alongside ILLUM\_INTRUCK, a new multi-object industrial benchmark dataset. Through 18 controlled experiments utilizing a state-of-the-art YOLOv12 framework, we demonstrate that complex, indirect lighting configurations paired with domain-relevant background variability significantly increase visual cue richness. Our quantitative findings show that avoiding direct specular peaks preserves crucial surface textures, mitigates the domain gap, reduces false positives, and accelerates model convergence compared to using conventional direct-light synthetic data. Ultimately, we provide actionable virtual scene design guidelines to maximize object detection robustness in industrial automation. --- ### What are Key Factors for Updates in RL for LLM Reasoning? Authors: Peidong Wang, Demi Wang, Xufang Luo, Jiahang Xu, Xiaocui Yang, Shi Feng, Yuqing Yang, Dongsheng Li Published: 2026-06-21 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.22570v2 Reinforcement Learning from Verifiable Rewards (RLVR) has emerged as a promising framework for enhancing the reasoning ability of large language models. However, much of the existing work is guided by heuristic intuition, leading to divergent algorithmic choices, even contradictory ones that nevertheless report empirical gains. To better understand this phenomenon, we conduct a theoretical analysis of RLVR updates. Our study reveals that differences in off-policy degree, determined by the number of gradient steps per rollout, substantially affect the distribution of importance sampling ratios and their clipping behavior, thereby altering which tokens dominate the update. Building on this insight, we characterize gradient expectation as the central quantity governing update dynamics and analyze the roles of token probability, advantage, and importance sampling ratio. Motivated by these findings, we propose Adaptive Clip Policy Optimization (ACPO), which adjusts clipping boundaries across token groups according to the empirical variance of their importance sampling ratios. Experiments on 3B and 7B models across diverse reasoning benchmarks, spanning mathematical problem solving, tabular QA, and logic puzzles, demonstrate that ACPO outperforms strong baselines such as DAPO and CISPO. These results demonstrate that principled, analysis-driven approaches yield more robust and effective RLVR methods. Code is available in: https://github.com/Control-derek/ACPO ---