# Research Papers — 2026-05 · bundle page 17 of 326

> Full-text markdown of papers #801–850 of 16266 published in 2026-05, newest first. AI-readable mirrors of the full 16266-paper 2026-05 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-05/page/17
[← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/16.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-05/page/18.md)

### FLaG: Fine-Grained Latent Grouping for Hallucination Detection

Authors: Wentao Ye, Liyao Li, Zhiqing Xiao, Muzhi Zhu, Jiaqi Hu, Zhanming Shen, Xiaomeng Hu, Sean Du, Haobo Wang
Published: 2026-05-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.00301v1

Hallucinations in large language models (LLMs) arise from heterogeneous failure mechanisms, making reliable detection difficult for any single global uncertainty score. In this work, we formulate hallucination detection as a mechanism-aware evidence aggregation problem, where diverse representation- and token-level signals must be interpreted under multiple latent explanations. We propose FLaG, a lightweight hallucination detection framework that models correctness through a set of latent evidence groups. Each instance is softly associated with multiple groups via an energy-based routing mechanism, and group-conditional reliability signals are combined through a principled log-marginal aggregation. This design enables FLaG to capture heterogeneous hallucination patterns while remaining invariant to decision thresholds and evaluation metrics. The framework operates as a frozen-model head, requires no modification to the underlying language model, and incurs minimal computational overhead. We further provide a theoretical perspective that connects FLaG to optimal evidence aggregation under heterogeneous error mechanisms, showing that the Bayes-optimal test statistic necessarily admits a log-marginal form and that FLaG constitutes a tractable approximation with a controllable error bound. Extensive experiments across multiple benchmarks and LLM backbones demonstrate that FLaG consistently achieves SOTA performance, while exhibiting robust transfer across datasets and models, and remaining effective under limited supervision.

---

### Real2SAM2Real: Generative 3D Caches as Complementary Context for Video Diffusion

Authors: Jiayi Wu, Haoming Cai, Cornelia Fermuller, Christopher Metzler, Yiannis Aloimonos
Published: 2026-05-29
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2606.00299v1

While Video Diffusion Models (VDMs) excel at synthesizing high-fidelity videos, enabling precise camera and scene control remains challenging. Existing methods predominantly rely on implicit diffusion priors to generate unobserved regions, inevitably leading to structural collapse during high-dynamic movements or complex occlusions. To address this challenge, we propose Real2SAM2Real, a framework that leverages 3D lifting models (e.g., SAM3D) to extract an explicitly editable 3D cache, serving as a robust geometric scaffold for the VDM. By capturing the entire 3D volume of foreground entities rather than just their visible shells, this cache injects holistic spatial priors into the VDM, providing dependable 3D-aware guidance for complex scene dynamics. To effectively leverage this 3D guidance while preserving pre-trained priors, we design a Soft Spatial-Aligned Injection mechanism alongside a minimally invasive fine-tuning strategy tailored for VDMs. Furthermore, we employ masked normal maps as a cross-modal bridge to construct a 3D-free data curation and perturbation pipeline. Extensive experiments demonstrate that Real2SAM2Real enables precise, decoupled control over both camera trajectories and multi-entity motions. By utilizing the complementary context from generative 3D caches, our framework overcomes typical breakdowns caused by over-reliance on diffusion priors, maintaining exceptional spatiotemporal consistency under large camera shifts and severe occlusions. Crucially, by decoupling geometry from appearance, our VDM-tailored 3D cache eradicates perspective ambiguities caused by structural holes and erroneous facades, as well as misleading cues from reflections and refractions. Project website is available at https://jiayi-wu-leo.github.io/real2sam2real

---

### Symmetric Hermite quadrature-based balanced truncation for learning linear dynamical systems from derivative data

Authors: Sean Reiter, Steffen W. R. Werner
Published: 2026-05-29
Categories: math.NA, cs.LG, eess.SY, math.DS, math.OC
arXiv: https://arxiv.org/abs/2606.00298v1

Data-driven reduced-order modeling is an essential component in the computer-aided design of control systems. In this work, we present a novel symmetric Hermite formulation of the quadrature-based balanced truncation algorithm that constructs linear reduced-order models from evaluations of the full-order system's transfer function and its derivative. Significantly, the Hermite formulation preserves desirable qualitative properties of the system used to generate the data, such as state-space Hermiticity and, consequently, asymptotic stability.

---

### Is Zero-Shot Super-Resolution Possible in Operator Learning?

Authors: Unique Subedi, Ambuj Tewari
Published: 2026-05-29
Categories: stat.ML, cs.LG, math.AP
arXiv: https://arxiv.org/abs/2606.00296v1

Neural operators are often reported to exhibit zero-shot super-resolution, a phenomenon in which a model trained on coarse grids produces accurate predictions on finer testing grids without additional retraining. Despite strong empirical evidence, the theoretical foundations of this phenomenon remain unclear. In this work, we provide a systematic theoretical study of zero-shot super-resolution in operator learning. We first show that zero-shot super-resolution can be information-theoretically impossible even in benign settings such as when the input functions are available over the entire continuum and the ground truth is a simple rank-one linear operator. We then identify H{\" o}lder smoothness of the output functions as a sufficient condition for zero-shot super-resolution and derive corresponding generalization bounds. Finally, we also validate the identified failure modes through experimental results.

---

### Adaptive Order Policies for Masked Diffusion

Authors: Jama Hussein Mohamud, Mohsin Hasan, Mirco Ravanelli, Yoshua Bengio
Published: 2026-05-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.00295v2

Masked diffusion models have seen great success in capturing data distributions over discrete sequences in domains such as text and proteins. These models generate data by iteratively unmasking tokens starting from a fully masked sequence, with the unmasking order typically chosen at random or using a heuristic based on denoiser probabilities. In this work, we propose a scheme for learning the unmasking order using an additional lightweight policy network on top of a diffusion model. Our proposed loss reweights terms in the masked diffusion loss according to policy probabilities, and results in a policy that prefers positions where the denoiser is more likely to be correct. We study this loss in two settings: (i) training solely the policy while using a frozen pre-trained denoiser, and (ii) training the policy and denoiser jointly with the weighted loss to allow for mutual adaptation. We demonstrate that our approach outperforms common heuristics on problems that are sensitive to token ordering, such as combinatorial problems, proteins as well as various coding and language tasks.

---

### Adaptive Order Policies for Masked Diffusion

Authors: Jama Hussein Mohamud, Mohsin Hasan, Mirco Ravanelli, Yoshua Bengio
Published: 2026-05-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.00295v1

Masked diffusion models have seen great success in capturing data distributions over discrete sequences in domains such as text and proteins. These models generate data by iteratively unmasking tokens starting from a fully masked sequence, with the unmasking order typically chosen at random or using a heuristic based on denoiser probabilities. In this work, we propose a scheme for learning the unmasking order using an additional lightweight policy network on top of a diffusion model. Our proposed loss reweights terms in the masked diffusion loss according to policy probabilities, and results in a policy that prefers positions where the denoiser is more likely to be correct. We study this loss in two settings: (i) training solely the policy while using a frozen pre-trained denoiser, and (ii) training the policy and denoiser jointly with the weighted loss to allow for mutual adaptation. We demonstrate that our approach outperforms common heuristics on problems that are sensitive to token ordering, such as combinatorial tasks and proteins.

---

### Uncovering Temporal Framing in the News

Authors: Tarek Mahmoud, Veronika Solopova, Premtim Sahitaj, Ariana Sahitaj, Max Upravitelev, Mervat Abassy, Hana Fatima Shaikh, Neda Foroutan, Vera Schmitt, Preslav Nakov
Published: 2026-05-29
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.00294v1

Temporal language does more than place events on a timeline. In news discourse, references to the past, present, and future can function as rhetorical devices that shape interpretation and persuasion. Here, we study temporal framing, defined as the persuasive use of time-related language to structure meaning rather than to report chronology. We propose a taxonomy of eight temporal frames grounded in prior work on temporality and framing, and we realize it through expert annotation of a multilingual news corpus. The resulting dataset includes 458 English and German news articles, with over 2K temporally framed sentences and approximately 3K temporal framing annotations identified from a corpus of more than 20K sentences. We analyze frame prevalence, co-occurrence patterns, and lexical cues, and evaluate temporal framing detection using supervised fine-tuning and zero-shot classification. Our experiments show that temporal framing is learnable at the sentence level, with supervised models substantially outperforming zero-shot approaches. We publicly release the corpus to support future research on temporal framing: https://mbzuai-nlp.github.io/temporal-framing/.

---

### Accurate Large-sample Uncertainty Quantification using Stochastic Gradient Markov Chain Monte Carlo

Authors: Yu Wang, Jie Ding, Jonathan H. Huggins
Published: 2026-05-29
Categories: cs.LG, stat.ME, stat.ML
arXiv: https://arxiv.org/abs/2606.00293v1

Tuning algorithms such as stochastic gradient descent (SGD) and stochastic gradient Langevin dynamics (SGLD) for approximate sampling and uncertainty quantification remains challenging, particularly in the practically relevant settings when the batch size is large or the model is misspecified. Existing theory that provides tuning guidance relies on continuous-time limits or strong statistical assumptions, which can become quantitatively inaccurate in these regimes. We address these shortcomings by proposing new discrete-time approximations to SG(L)D with and without momentum, which enables accurate predictions of the stationary covariance, iterate average covariance, and integrated autocorrelation time. Moreover, we prove quantitative, non-asymptotic error bounds showing that these estimates are sufficiently accurate for practical tuning and uncertainty quantification. Numerical experiments demonstrate that our theory yields improved tuning guidance across a range of models and data-generating distributions where existing approaches fail, including when using the $β$-divergence rather than log-loss to obtain statistically robust inferences.

---

### The Representation-Rationalizability Tradeoff in Reward Learning

Authors: Jing Dong, Yaoliang Yu, Pascal Pourpart
Published: 2026-05-29
Categories: cs.GT, cs.LG
arXiv: https://arxiv.org/abs/2606.00291v1

In RLHF, each training example contains a prompt $x$ and two candidate responses $y,y'$, and annotators provide pairwise preferences between these responses. The learning problem is to convert these heterogeneous pairwise judgments into a single scalar reward $r(x,y)$ that measures response quality for each prompt. Classical social choice implies an impossibility because heterogeneous annotator samples can induce pooled preferences with Condorcet cycles, so no scalar reward can evaluate all compared response pairs consistently. A growing literature analyzes RLHF as a social-choice problem, but usually assumes a fixed finite set of alternatives, i.e., a pre-enumerated finite set of candidate responses for each prompt. Modern pipelines instead score responses through a learned representation $φ(x,y)$ before a scalar head, so $φ$ determines which responses are treated as distinguishable alternatives and which comparisons are visible to the reward model. Once this embedding is part of the problem, the impossibility results from social choice theory become a tradeoff. We show that the excess cross-entropy loss of any reward built on $φ$ decomposes exactly into a representational term, which a richer $φ$ shrinks, and an aggregation term, which a richer $φ$ enlarges by exposing more comparisons that no scalar can rank consistently. The same results extend to direct preference optimization (DPO), and jointly training the embedding and the reward cannot guarantee to recover the sweet spot of this tradeoff. Experiments on synthetic data and real preference datasets corroborate our results.

---

### Inner Product Aware Quantization: Provably Fast, Accurate, and Adaptive Algorithms

Authors: Nathan White, Krish Singal
Published: 2026-05-29
Categories: cs.LG, cs.DS
arXiv: https://arxiv.org/abs/2606.00289v1

Quantization is a fundamental tool used to compress datasets, neural network weights, and memory usage in a range of computational tasks. Many downstream applications of vector quantization perform inner products with arbitrary inputs. This motivates the study of inner product aware quantization schemes that approximately preserve inner products with unseen vectors -- in contrast to simply minimizing the mean-squared error. In this work, we formulate objectives that capture natural desiderata and develop adaptive and unbiased quantization methods that approximately preserve inner products with worst-case and average-case inputs. An analysis of these objectives shows a tight connection with the well-studied notion of Adaptive Stochastic Quantization (ASQ). We develop provably fast exact and approximate algorithms for our objectives. Our theoretical results inspire efficient practical algorithms that perform well across a variety of workload distributions. They also lead to practical algorithms for standard ASQ which are 2-10$\times$ faster than prior state-of-the-art methods while maintaining quality. These theoretical and empirical results contribute towards making adaptive quantization techniques more efficient and tractable in practical settings.

---

### Model-Native Computing Architecture: Envisioning Future System Architecture Through the Lens of Computer Architecture

Authors: Hai Lin, Hoilam Pao, Shaoxiong Zhan, Hai-Tao Zheng
Published: 2026-05-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.00288v3

Large language models are undergoing a transition from model technology to system technology. Engineering challenges like cache reuse, context capacity, agent scheduling, and permission control resemble classical computer systems problems. This raises a question: if we treat the LLM as a CPU, KV cache as processor cache, context window as main memory, and agent framework as an operating system, can decades of computer architecture wisdom guide next generation model native systems? This paper pursues this analogy as a visionary survey. We map computer architecture concepts onto the emerging model native stack, survey literature across LLM as OS, memory management, agent frameworks, tool protocols, multi agent coordination, cognitive architectures, and safety governance, finding that each addresses a different layer without a unifying model. We propose the Intelligent Computing Architecture (ICA): six functional layers with interface contracts and design axioms. We resolve the tension over whether the LLM resembles a CPU or OS via a dual plane architecture a probabilistic execution plane (what can be computed) and a deterministic control plane (what should be computed), with every layer passing through as a graded crossover. We propose three Amdahl style design heuristics Semantic Locality, Context Budget, and Agent Speedup as organizing back of envelope models, illustrate their parameter ranges with published data, and identify predictive validation as the principal open task. We articulate analogy boundaries, note differences between silicon and model era architectures, and propose a research roadmap. This is a conceptual and survey contribution with no new experimental results.

---

### Model-Based Quality Assessment for Massively Multilingual Parallel Data

Authors: Abdelaziz M. A. Ibrahim, Zihao Li, Jörg Tiedemann, Shaoxiong Ji
Published: 2026-05-29
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.00285v1

Large-scale multilingual bitext often contains two distinct problems: non-parallel sentence pairs and low-quality translations. We decompose model-based assessment for such data into two independent components: parallelism assessment with multilingual embeddings and reference-free quality estimation (QE). For parallelism, we benchmark four embedding models on FLORES-200 and BOUQuET retrieval tasks, covering 6,654 source--target directions in our target language-pair inventory. For QE, we evaluate nine reference-free evaluators on professional FLORES-200 translations across 41,412 ordered source--target directions. Results show that no model is universally reliable across translation directions. Naive QE ensembles dilute strong model signals, while documented target-language coverage is strongly associated with higher QE scores. Overall, these findings suggest that multilingual parallel-data assessment is best approached as a direction-aware routing and calibration problem, where no single universal metric is expected to suffice across all languages.

---

### Parameter Alignment Mitigates Catastrophic Forgetting in Multilingual Expert Language Models

Authors: Sanchit Ahuja, Terra Blevins
Published: 2026-05-29
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.00284v1

While continual pretraining~(CPT) is a practical way to extend large language models to new languages, naïve finetuning on targeted data erodes existing capabilities through catastrophic forgetting. Organizing training around language families reduces cross-language interference but cannot alone prevent forgetting of the general knowledge needed for downstream tasks. We link this forgetting to parameter drift in multilingual CPT and present a suite of five layer-aware parameter alignment strategies: hard layer freezing, soft regularization, post-hoc weight reversion, and model merging. We systematically compare our alignment strategies against two unregularized CPT baselines on benchmarks spanning 32 training languages from five language families, plus held-out languages, across four evaluation axes: perplexity, reading comprehension, physical reasoning, and translation. Parameter alignment substantially reduces forgetting at minimal cost to language acquisition: layer freezing and regularization best preserve comprehension, whereas post-hoc reversion yields the strongest translation gains. Together, these results map the acquisition--forgetting frontier for family-expert CPT and offer practical deployment guidelines pairing each strategy to the tasks it best serves.

---

### Synthetic Data from Cross-Domain Events for Large-Scale Recommendation Systems

Authors: Xiangyu Wang, Yawen He, Shivendra Pratap Singh, Han Huang, Mengtong Hu, Sharath Ciddu, Yi-Hsuan Hsieh, Erik Groving, Yi Ding, Jieming Di, Tony Wang, Min Yun et al.
Published: 2026-05-29
Categories: cs.IR, cs.AI
arXiv: https://arxiv.org/abs/2606.00282v1

Large-scale recommendation systems operate across diverse domains, yet they face the challenges of data sparsity and noisy implicit feedback. Traditional approaches mitigate this via model-specific knowledge distillation from source domains to a target domain. Inspired by the transformative success of synthetic data generation in large language models (LLMs), we introduce Synthetic Cross-domain Augmentation and Learning for Recommendation (SCALR), a framework that generates synthetic user-item interaction events for a target recommendation domain by leveraging observed events from a source domain. SCALR decomposes cross-domain learning into two modular stages. First, it translates observed user events in source domains by framing event generation as estimating the likelihood that a user would interact with a target-domain item, conditioned on their observed interactions in a source domain. Second, downstream models train on these synthetic events as cross-domain learning objectives, where the synthetic events augment the target domain's training data in a model-agnostic manner. Our approach yields statistically significant improvements in online A/B tests on an industrial recommendation platform. To the best of our knowledge, this is among the first works to explicitly frame cross-domain event transfer as synthetic data generation for recommendation systems.

---

### Flow Matching for Convective-Scale Precipitation Downscaling

Authors: Tom Wetherell
Published: 2026-05-29
Categories: physics.ao-ph, cs.LG
arXiv: https://arxiv.org/abs/2606.00281v1

Generative machine learning is an increasingly important complement to dynamical downscaling for producing high-resolution precipitation projections, with diffusion models currently the leading approach. Flow matching is a related generative framework that has recently achieved strong results across image, video and other domains, and shown early promise for downscaling. We train a flow matching model to map daily precipitation from 8 km to 2 km over a convective-scale domain centred on Singapore, and benchmark it against CPMGEM, a score-based diffusion model. Flow matching achieves consistently better spatial skill: higher fractions skill score at every precipitation threshold and neighbourhood scale tested, and tighter structure and amplitude components of the SAL score with comparable location skill. However, flow matching underestimates the upper tail of the precipitation distribution, resulting in a dry bias in the climatological mean. These results suggest that flow matching is a competitive generative framework for convective-scale precipitation downscaling, particularly well suited to capturing spatial structure.

---

### Bit-Exact AI Inference Verification Without Performance Tradeoffs

Authors: Naci Cankaya
Published: 2026-05-29
Categories: cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2606.00279v2

Verifying claims about AI workloads is a prerequisite for credible AI governance of covert adversaries (who comply with monitoring only when detection likelihood is high), yet the apparent non-determinism of GPU floating-point arithmetic forces auditors to accept approximate output matches. Covert adversaries can exploit unverifiable degrees of freedom in monitored computation. Attack vectors include steganography, unreported modification of inference software, and covert computation via unreported batch elements. Empirically, we analyze how modern inference engines (vLLM, HF transformers) produce deterministic but non-invariant outputs, without needing to set performance-compromising determinism flags, if the right information is available for re-computation and no atomic functions are called in the backend. We demonstrate that such bitwise-precise re-computation does not require access to identical hardware, via a software-only emulation of LLM inference across multiple NVIDIA GPU variants. Thus, accumulated rounding errors can be an auditable signature of the software and hardware setup used for inference, instead of a constraint on verifiability.

---

### Evaluating Bivariate Causal Statements Based on Mutual Compatibility

Authors: Erik Jahn, Dominik Janzing
Published: 2026-05-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.00278v1

For many real-world systems, causal ground truth is difficult to obtain, making claims about causal effects hard to assess. We develop methods for evaluating collections of $\binom{n}{2}$ bivariate causal statements over a set of $n$ variables. In the setting of acyclic linear statements, any such collection can be extended to a unique multivariate causal model, but we argue that this induced model is implausible if it imposes substantial additional confounding to explain observed correlations. We introduce a compatibility score that quantifies this notion of plausibility, notably without relying on the faithfulness assumption. Additionally, we define an incompatibility score for purely graphical bivariate causal statements, based on global consistency constraints that are derived from acyclicity and faithfulness assumptions. We give theoretical and empirical evidence that both scores can successfully distinguish correct from incorrect causal statements in generic settings. Moreover, we demonstrate the practical applicability of our methods by analyzing causal claims made by large language models. Our work aims to provide a foundation for assessing the reliability of causal information derived from human experts or artificial intelligence in settings where alternative forms of validation are unavailable.

---

### Hyperbolic and Evidence-Prioritized Experts for Large Vision-Language Models

Authors: Zijie Zhou, Dandan Zhu, Hangxiangpan Wang, Heng Zhang, Huishen Jiao, Yi Zhao
Published: 2026-05-29
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2606.00275v1

Large Vision-Language Models (LVLMs) have demonstrated impressive performance on multimodal tasks through scaled architectures and extensive training. Recent studies introduce Mixture of Experts (MoE) into LVLMs for improved computational efficiency. However, existing MoE approaches treat visual and linguistic modalities with symmetric architectures, overlooking the inherent asymmetry in how these two modalities are processed. This asymmetry causes two critical issues. First, text and vision form hierarchical rather than parallel relationships, as text queries typically describe partial aspects of complete visual scenes. Euclidean expert space struggles to encode such containment structures. Second, language experts in deeper layers progressively shift from evidence-based processing to parametric memory dependence, losing grounding in the provided visual and linguistic information. To address these issues, we propose AsyMoE, a novel architecture that explicitly models this asymmetry through three specialized expert groups. Intra-modality experts handle modality-specific processing. Hyperbolic inter-modality experts capture hierarchical cross-modal relationships through negative curvature geometry. Evidence-priority language experts suppress parametric memory activation and maintain contextual grounding throughout network depth. Extensive experiments demonstrate that AsyMoE achieves consistent improvements over baseline methods, with average gains of 1.5\% over MoE variants and up to 3.8\% on hallucination-sensitive tasks. AsyMoE activates 25.45\% fewer parameters compared to dense models.

---

### On Wednesdays, We Ask Questions: Optimizing "Active Listening" in Automated Legal Triage and Referral

Authors: Quinten Steenhuis, Jacqueline Harvey
Published: 2026-05-29
Categories: cs.AI, cs.CL, cs.CY
arXiv: https://arxiv.org/abs/2606.00272v1

The FETCH classifier generates follow-up questions to help refine the best match for the applicant's legal problem, using a low-cost ensemble of LLMs. In this paper, we describe an expert attorney and LLM-assisted evaluation of the follow-up question approach in FETCH and show that while low-cost LLMs perform well at classification tasks, generating high-quality plain-language questions in this setting appears to require a more sophisticated and higher-cost model. Through discussion with legal intake workers, we propose a rubric for the evaluation of legal intake classification questions, and we find that prompt engineering alone is not enough to improve question quality for intake purposes. We also find that LLM-as-judge and human ratings diverge. We demonstrate that with the addition of a single high-cost model, GPT-5, the classifier can elicit relevant information from applicants for legal help, and that the questions lead to more accurate performance at classification tasks. We also find uneven fact elicitation across different categories, including domestic violence, at odds with family law screening protocols, suggesting the value of including dedicated screening panels for certain areas of law.

---

### Robust Shielding for Safe Reinforcement Learning

Authors: Edwin Hamel-De le Court, Thom Badings, Alessandro Abate, Francesco Belardinelli, Francesco Fabiano
Published: 2026-05-29
Categories: cs.AI, cs.LG, cs.LO
arXiv: https://arxiv.org/abs/2606.00270v1

Shielding is an effective approach to formally guarantee the safety of reinforcement learning agents in Markov decision processes (MDPs). However, existing shielding techniques typically assume knowledge of the safety-relevant transition dynamics - a requirement that is seldom met in practice. To address this limitation, we introduce a novel shielding framework for robust MDPs (RMDPs), i.e., MDPs with sets of transition probabilities. We define safety as the satisfaction of a linear temporal logic (LTL) formula with a certain threshold probability under the worst-case transition probabilities of the RMDP. We prove that our shielding framework is both sound and optimal for the RMDP: every policy admissible by the shield is safe, and conversely, every safe RMDP policy is admissible by the shield. We combine our approach with existing sampling methods for learning transition probabilities of MDPs with probably approximately correct (PAC) guarantees. This combination enables the construction of shields for MDPs that, with high confidence, guarantee safety while remaining minimally restrictive. Our experiments show that our shields for learned RMDPs guarantee safety in unknown MDPs while recovering strong expected return as the number of samples increases.

---

### Closed-Loop Neural Activation Control in Vision-Language-Action Models

Authors: Abhijith Babu, Ramneet Kaur, Nathaniel D. Bastian, Olivera Kotevska, Susmit Jha, Yanzhao Wu, Sumit Kumar Jha, Anirban Roy
Published: 2026-05-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.00269v1

Vision-Language-Action (VLA) models can be steered at test time by intervening on semantically meaningful internal directions, but existing methods use a fixed steering coefficient, effectively operating in open loop. This is poorly suited to embodied control, where task state and concept error evolve over time, often causing overcorrection, oscillation, and reduced task success, especially for temporal behaviors such as speed and smoothness. We propose CTRL-STEER, a closed-loop framework that replaces static intervention strength with adaptive, time-varying control signals. The key idea is to decouple representation from regulation: rather than assuming temporal concepts are directly controlled by individual neurons, we steer along motion-aligned residual directions while a feedback controller adjusts intervention magnitude online. We instantiate this framework with both PID and reinforcement learning based controllers. Experiments with a fine-tuned OpenVLA policy on four LIBERO task suites show that CTRL-STEER achieves more stable concept regulation and a better steering-task success trade-off than fixed-coefficient baselines, without modifying or retraining the base model.

---

### StressDream: Steering Video World Models for Robust Policy Evaluation and Improvement

Authors: Junwon Seo, Sushant Veer, Ran Tian, Wenhao Ding, Apoorva Sharma, Karen Leung, Edward Schmerling, Marco Pavone, Andrea Bajcsy
Published: 2026-05-29
Categories: cs.CV, cs.AI, cs.LG, cs.RO
arXiv: https://arxiv.org/abs/2606.00267v1

Video world models (WMs) have shown promise for policy evaluation and improvement by imagining realistic future observations conditioned on ego-robot actions. While WMs can model distributions over futures, policy evaluation and improvement typically rely on nominal imaginations, which can miss high-impact outcomes of robot actions unless prohibitively many samples are drawn. To enable robust policy evaluation and improvement over WM imaginations, we propose StressDream, which steers imaginations toward high-impact yet plausible outcomes specified at inference time by optimizing the initial noise of diffusion-based WMs. However, optimizing high-dimensional noise is challenging: the optimization must reason about nuanced, scene-dependent target events in generated videos while avoiding out-of-distribution (OOD) noise that yields implausible imaginations. We address this with two complementary objectives: a semantic objective with a Vision-Language Model that provides informative gradients by reasoning about the generated video, and a plausibility objective that prevents the optimized noise from drifting OOD. With state-of-the-art video world models for autonomous driving and robotic manipulation, we show that StressDream effectively steers imaginations toward high-impact yet plausible outcomes specified by text at inference time, such as task failures, enabling robust policy evaluation and improvement by identifying actions whose plausible futures include undesirable outcomes. Video results are available at https://junwon.me/StressDream/.

---

### KISS: Keeping it Simple and Slotted when Learning to Communicate over Wireless

Authors: Kamil Szczech, Maksymilian Wojnar, Krzysztof Rusek, Katarzyna Kosek-Szott, Szymon Szott
Published: 2026-05-29
Categories: cs.NI, cs.LG
arXiv: https://arxiv.org/abs/2606.00266v1

A long-standing challenge in distributed wireless systems is ensuring efficient and fair random channel access. Existing solutions often address specific constraints related to timing, periodicity, or centralization, but they typically rely on fixed heuristics. Motivated by recent advances in machine learning (ML), we investigate whether ML agents can autonomously learn efficient and fair access strategies, and whether such learning can offer new insights into medium access control (MAC) design. Rather than proposing a deployable protocol, our aim is to examine whether decentralized learning can rediscover or approximate theoretically efficient random-access mechanisms under minimal assumptions. To this end, we deploy an off-policy Double Deep Q-Network (DDQN) with Bayesian inference to train agents operating over a slotted channel. The resulting method is fully online (no pre-training), fully distributed (independent multi-agent learners), stochastic (non-periodic), and requires no coordination or explicit communication. Extensive simulations show that the learned strategy adapts to varying network conditions and achieves near-theoretical efficiency while maintaining fairness. Ablation studies further reveal that the learned behavior resembles slotted ALOHA with a dynamically adjusted transmission probability, leading us to refer to the method as KISS: Keeping It Simple and Slotted.

---

### Out-of-Distribution generalization of quantile regression with heavy tailed inputs: an SVM approach

Authors: Baptiste Leroux, Clément Dombry, Anne Sabourin
Published: 2026-05-29
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2606.00265v1

We study quantile regression in an extrapolation regime where the covariate takes unusually large values. Under regular variation assumptions, extreme observations can be effectively characterized through their angular components, enabling learning strategies that focus on the angle of the most extreme observations. This approach is formalized through the minimization of an asymptotic conditional risk that localizes learning in the tail of the covariate distribution. We propose a novel Support Vector Machine (SVM) framework for extreme quantile regression, leveraging reproducing kernel Hilbert spaces to handle high-dimensional and nonlinear settings. Our method also accommodates unbounded response variables and avoids restrictive transformations. We establish finite-sample learning guarantees under mild regularity assumptions. The proposed framework unifies ideas from statistical learning and multivariate extremes, providing a tractable and theoretically grounded approach to extrapolation. We complement our theoretical findings with an empirical study on river flow data from the Danube, demonstrating the practical relevance of our methods.

---

### ReFLEX: Length-Generalizable CSI Denoising for MIMO-OFDM via Relative-Frequency Bias

Authors: Zhibin Zhang, Robert Potekhin, Ziwei Wan, Vladimir Lyashev, Zhen Gao
Published: 2026-05-29
Categories: eess.SP, cs.LG
arXiv: https://arxiv.org/abs/2606.00263v1

This letter studies CSI denoising for MIMO--OFDM with variable NR resource block (RB) allocations. ReFLEX is a length-generalizable Transformer whose frequency attention uses a relative-frequency position bias (RFPB) generated from subcarrier offsets. A single checkpoint handles unseen RB lengths and can be applied to sparse DM-RS observations in the tested RB5/RB10 PUSCH setup without retraining. In a 3GPP~TR~38.901 UMa NLOS channel, ReFLEX achieves about $-9.6$~dB NMSE on unseen RB lengths. In NR PUSCH/UL-SCH simulations, ReFLEX denoising followed by time-frequency interpolation reduces the 10\% BLER threshold by about 2--3~dB.

---

### When Softmax Fails at the Top: Extreme Value Corrections for InfoNCE

Authors: Melihcan Erol, Suat Evren, Oktay Ozel, Alexander Morgan, Jongha Jon Ryu, Lizhong Zheng
Published: 2026-05-29
Categories: cs.LG, cs.AI, stat.AP, stat.ML
arXiv: https://arxiv.org/abs/2606.00262v1

InfoNCE is the standard contrastive learning objective, but its softmax form is not only a computational convenience: it also encodes a statistical assumption about how the top-scoring example is selected. Using extreme value theory, we show that this assumption is often misaligned with the normalized embedding setting used in modern contrastive learning. Motivated by this mismatch, we propose \textsc{WEINCE}, a simple modification of InfoNCE that uses anchor-wise online batch statistics to blend the usual softmax logits with an endpoint shortfall correction, adding no trainable parameters. Across five vision benchmarks, \textsc{WEINCE} yields consistent improvements in frozen-feature evaluation. These results show that a more faithful statistical treatment of hard negatives can improve contrastive objectives.

---

### The Harsh Truth: Segment-Level Analysis of Harsh Driving Events in Milan Using Large-Scale Telematics, Street Networks, and Google Street View

Authors: Andrea La Grotteria, Paolo Santi, Titus Venverloo, Umberto Fugiglando, Carlo Ratti
Published: 2026-05-29
Categories: cs.CV, physics.soc-ph
arXiv: https://arxiv.org/abs/2606.00261v1

Police-reported crash statistics remain the standard input for urban road-safety assessment, but their incompleteness and reporting lag limit their usefulness for timely, fine-grained intervention design. Harsh acceleration and braking events are widely used as surrogate safety indicators, but have so far been studied only in comparatively small urban samples. This study analyses harsh events across the urban road network of Milan, combining high-resolution telematics from more than 4.2 million vehicles equipped with On-Board Units, segment-level traffic metrics from TomTom, street-network and infrastructure attributes from OpenStreetMap, and visual streetscape features extracted from Google Street View via semantic segmentation using a OneFormer model. We employ an analytical framework combining non-parametric Mann--Whitney U tests of segment-feature distributions between high- and low-harshness groups with supervised machine-learning regressors. We find that, once exposure is controlled for, wider carriageways, crossings and transit stops, and more open visual fields (higher sky- and road-pixel proportions) are associated with higher harsh-event intensity, while denser built frontage is associated with lower intensity. Finally, the cycling-infrastructure case study identifies a gradient in harsh-event intensity across facility types: markings-only cycle lanes are associated with a 19.5% higher harshness score, and mixed-traffic configurations with an 11.5% higher score, relative to physically separated cycle paths, conditional on the included controls. These results support context-specific rather than uniform urban-safety interventions and illustrate how large-scale telematics combined with open geospatial and visual data can inform Vision Zero decision-making at the metropolitan scale.

---

### LastAct: Trajectory-Guided Latest-Activity Localization for Real-Time Smart-Home Activity Recognition

Authors: Zishuai Liu, Ruili Fang, Jin Lu, Fei Dou
Published: 2026-05-29
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2606.00260v3

Human Activity Recognition (HAR) from ambient sensors enables smart-home applications such as health monitoring and assisted living. In realistic deployments, however, sensor events arrive as a continuous stream and activity boundaries are unknown. Sliding-window inference therefore produces many windows that straddle transitions and contain mixed activities, creating boundary contamination that violates the pre-segmented instance assumption used by most benchmarks and models. Moreover, many pipelines under-use spatial context by treating sensor IDs as independent tokens. We present LastAct, a trajectory-centric framework for streaming smart-home HAR that targets the most recent activity under mixed windows while explicitly modeling spatial structure. LastAct projects sensor events onto the home floorplan to form a layout-aligned trajectory image sequence that preserves spatial continuity. A lightweight gate identifies contaminated windows, and a boundary localizer estimates the last transition to enable boundary-guided masking that emphasizes post-boundary evidence and suppresses stale context. For efficiency, we reuse a precomputed layout-aligned template cache to avoid repeated rendering. Empirically, across four public smart-home datasets under near-realistic mixed-activity protocols, LastAct achieves competitive or superior performance on pure windows and yields substantial Macro-F1 gains on cross/mixed windows, demonstrating improved robustness under near-realistic sliding-window regimes.

---

### ARCA: Adapter-Residual Credit Assignment When Token Signals Degenerate

Authors: Rodney Lafuente-Mercado
Published: 2026-05-29
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.00257v1

Token-level credit assignment for language-model reinforcement learning is usually formulated as if the policy were fully trainable, while practical LLM-RL pipelines often rely on parameter-efficient fine-tuning, especially LoRA. We argue that this separation hides a structural failure mode. Under LoRA, the policy is restricted to a low-rank neighborhood of the reference model, so the per-token output-distribution differences used by common intrinsic credit signals, surprisal, entropy reduction, and policy divergence, can become degenerate after within-trajectory normalization, either approaching uniform weights or concentrating on a small set of task-agnostic positions. We formalize this behavior and propose measuring it directly with concentration diagnostics such as weight Gini and effective-token ratio. We then introduce \emph{Adapter-Residual Credit Assignment} (ARCA), a lightweight alternative that derives token salience from the adapter's own hidden-state residual, $\|h^{\text{adapted}}_t - h^{\text{base}}_t\|_2$. ARCA asks where the adapter actually changes the model, rather than where the output distribution appears uncertain or shifted, and requires no learned reward model, value head, or tree construction. In a compact MATH/Qwen3-1.7B GRPO sweep, ARCA exhibits the predicted non-degenerate middle-regime credit distribution under matched rollout budgets and remains competitive with rank-matched baselines.

---

### Per-Group Error, Not Total MSE: Fine-Tuning Vision-Language-Action Models for 11-DoF Mobile Manipulation

Authors: Pau Montagut Bofi, Mario García Blasco, Tessa Pulli, Markus Vincze
Published: 2026-05-29
Categories: cs.RO, cs.LG
arXiv: https://arxiv.org/abs/2606.00253v1

Fine-tuning Vision-Language-Action (VLA) models for mobile manipulators with heterogeneous joint spaces can produce a counterintuitive result: the checkpoint with the lowest aggregate MSE is not the one that performs best on the real robot. We argue this is a predictable consequence of collapsing heterogeneous joint groups (arm, gripper, head, wheeled base) into a single metric, where easy-to-predict joints can mask joints that still fail. We fine-tune SmolVLA (450M, action-expert only) on the 11-DoF Toyota HSR and compare it against $π_{0.5}$ (3.3B), a stronger pretrained baseline. Per-group analysis exposes two patterns: in SmolVLA, the mobile base converges slowest and limits overall performance. In expert-only fine-tuning of $π_{0.5}$ (training only the action head, backbone frozen), total MSE drops below the baseline but arm accuracy degrades. On 60 real-robot trials (20 per model), $π_{0.5}$ 80k (4.0/4) significantly outperforms both fine-tuned variants (expert-only 3k: 3.75/4; HSR-SmolVLA: 3.5/4; Mann-Whitney $p \leq 0.010$), despite expert-only 3k having the lowest total MSE. This separation is most consistent with the offline arm-group error, not total MSE or base-group error. We conclude that per-group error is a more reliable signal than total MSE for checkpoint selection on robots with heterogeneous action spaces. Code: https://github.com/paumontagut/per-group-mse-vla

---

### HOIST: Humanoid Optimization with Imitation and Sample-efficient Tuning for Manipulating Suspended Loads

Authors: Songyang Liu, Shunyu Yao, Dingyuan Huang, Shuai Li
Published: 2026-05-29
Categories: cs.RO, cs.LG
arXiv: https://arxiv.org/abs/2606.00252v1

Manipulating suspended payloads with humanoid robots is challenging because the robot can only influence an underactuated, oscillatory load through whole-body motion and intermittent contact. Imitation learning provides safe initial behavior but does not directly optimize final placement, while reinforcement learning from scratch is unsafe and sample-inefficient on real humanoids. We present HOIST-Humanoid Optimized with Imitation and Sample-efficient Tuning for manipulating suspended loads. HOIST first finetunes a high-level vision-language-action (VLA) policy from virtual-reality (VR) teleoperation demonstrations and executes its commands through a whole-body controller. It then uses VLA rollouts and iterative batched RL to improve placement accuracy and stopping behavior. Experiments in simulation and on a real humanoid show that HOIST improves over imitation-only and additional-demonstration baselines; compared with pure VLA rollouts, HOIST reduces translational placement error by 19.9 cm and raw angular error by 3.56 degrees, demonstrating the potential of humanoids for underactuated material-handling tasks.

---

### Capability Self-Assessment in Large Language Models

Authors: Haoyan Yang, Reza Shirkavand, Yukai Jin, Jiawei Zhou, Shangqian Gao, Heng Huang
Published: 2026-05-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.00251v2

The ability to recognize one's own limitations and decide whether to solve a problem or seek help is fundamental to reliable intelligent systems. Yet modern large language models tend to overestimate their competence and attempt queries they cannot solve. We study how models can learn Capability Self-Assessment (CSA) while preserving their problem-solving ability. Because the same model both assesses and solves queries, self-assessment training can change the capability it must recognize. We systematically compare supervised fine-tuning (SFT) and reinforcement learning (RL), evaluating each trained model's self-assessment against its updated capability and measuring how well it retains its solving ability. Across model families, scales, and domains, RL provides the strongest assessment-retention trade-off. We also find that CSA transfers to the out-of-distribution settings we evaluate, and demonstrate its practical value for local-cloud routing at inference time and targeted data selection during training.

---

### Capability Self-Assessment: Teaching LLMs to Know Their Limits

Authors: Haoyan Yang, Reza Shirkavand, Yukai Jin, Jiawei Zhou, Shangqian Gao, Heng Huang
Published: 2026-05-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.00251v1

The ability to recognize one's own limitations and decide whether to solve a problem or delegate is fundamental for reliable intelligent systems. Yet we show that modern large language models systematically lack this ability: across diverse model families and scales, they overestimate their competence and attempt queries they cannot solve. We refer to this ability as Capability Self-Assessment (CSA) and formulate it as a policy-learning problem, aiming to improve self-assessment while preserving the model's original capabilities. Our results show that reinforcement learning teaches CSA effectively, significantly outperforming supervised fine-tuning while preserving original capabilities. In contrast, supervised fine-tuning severely degrades the capabilities the model is meant to assess. Moreover, learned self-assessment behavior generalizes well out of distribution, suggesting that CSA is a transferable model trait. Finally, CSA is practically useful: it improves local-cloud decision making at inference time and provides a signal for targeted data selection during training.

---

### Effects of Varying LLM Access on Essay Writing Behavior

Authors: Julia Christenson, Karin de Langis, Shirley Anugrah Hayati, Dongyeop Kang
Published: 2026-05-29
Categories: cs.CL, cs.AI, cs.HC
arXiv: https://arxiv.org/abs/2606.00250v1

Investigating the degree to which large language models (LLMs) affect teaching and learning in universities can help identify strategies for integrating LLMs in a way that supports, rather than undermines, student learning outcomes. This study examined how varying levels of LLM assistance affect writing performance, engagement, and perceived authorship. We report a pilot study in which 24 college students were randomly assigned to write a short essay with no LLM access, limited access (<=3 prompts, responses capped at 100 words), or unlimited access. Overall essay quality was statistically indistinguishable across groups. Yet writing behavior and perceived authorship diverged sharply: students with limited access reported higher ownership (62.5% would submit the essay as independent work, vs. 25% in the unlimited group), stronger organizational gains, and more strategic, revision-focused prompting. The unlimited group spent more time writing, produced essays more similar to LLM output, and reported reduced creative expression. Our findings suggest that constraining, rather than banning, LLM access may preserve authorship confidence while retaining the scaffolding benefits of AI assistance.

---

### Geodesic Flow Matching for Denoising High-Dimensional Structured Representations

Authors: Karim Habashy, Chris Eliasmith
Published: 2026-05-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.00248v1

Vector Symbolic Algebras (VSAs) enable robust neurosymbolic reasoning by encoding symbolic information into high-dimensional distributed representations. For continuous domains, Spatial Semantic Pointers (SSPs) extend this framework by mapping variables onto continuous toroidal manifolds. However, standard approaches like Flow Matching assume a flat Euclidean geometry, which fails to account for the geometric constraints imposed on valid SSP states. We demonstrate that this assumption fails for SSPs: Euclidean linear interpolants ``cut through" the manifold's interior, destroying the phase and magnitude structure required for accurate decoding. To resolve this, we employ Geodesic Flow Matching, adapting Riemannian transport dynamics to strictly restrict the denoising flow to the SSP toroidal manifold. We validate this approach in a Spiking Neural SLAM system, showing that manifold-aware cleanup stabilizes path integration against drift. The method achieves a 72\% reduction in tracking error and enables a 40\% increase in neural efficiency compared to competitive baselines. Code is available at https://github.com/kremHabashy/CleanupSSP .

---

### Dynamics and Representation Structure of Local Approximations to Gradient-Based Learning in Linear Recurrent Neural Networks

Authors: Ezekiel Williams, Alexandre Payeur, Guillaume Lajoie
Published: 2026-05-29
Categories: cs.NE, q-bio.NC, stat.ML
arXiv: https://arxiv.org/abs/2606.00243v1

Biological and neuromorphic recurrent neural networks (RNNs) are subject to spatial and temporal locality constraints on the information that can plausibly be used during learning. A common strategy to satisfy these constraints is to modify gradient descent by neglecting non-local terms to varying degrees, as in random feedback local online (RFLO) learning and truncated backpropagation through time (tBPTT). However, the learning dynamics of these algorithms, and how they compare with BPTT, remain poorly understood. We apply dynamical systems theory to data-aligned linear RNNs -- whose dynamics can be separated into orthogonal modes -- to compare stationary solutions, stability properties, and convergence rates, finding qualitatively distinct behaviour for RFLO versus BPTT and one-step tBPTT. We further observe that the solutions learned by RFLO are restricted to low-rank perturbations of initial parameters, a result which holds beyond the data-aligned setting. Our work provides analytical insight into how locality constraints shape learning dynamics, with implications for neuroscientific models of learning and alternative optimization approaches for RNNs.

---

### InfoAtlas: A Foundation Model for Zero-Shot Statistical Dependence Estimate

Authors: Zhengyang Hu, Yanzhi Chen, Hanxiang Ren, Qunsong Zeng, Youyi Zheng, Adrian Weller, Kaibin Huang, Yanchao Yang
Published: 2026-05-29
Categories: cs.LG, cs.AI, stat.ML
arXiv: https://arxiv.org/abs/2606.00241v2

Measuring statistical dependency between high-dimensional random variables is a fundamental task in data science and machine learning. Neural mutual information (MI) estimators offer a promising avenue, but they typically require costly iterative optimization for each new dataset, making them impractical for real-time applications. We present InfoAtlas, a foundation model-like architecture that eliminates this bottleneck by directly inferring MI in a single forward pass. Pretrained on large-scale synthetic data with rich dependence patterns, InfoAtlas learns to identify diverse dependence structures and predict MI directly from the dataset. Comprehensive experiments demonstrate that InfoAtlas matches state-of-the-art neural estimators in accuracy while achieving $100\times$ speedup, can flexibly handle varying dimensions and sample sizes through a single unified model, and generalizes effectively to complex, real-world scenarios. By reformulating MI estimation as an inference task, InfoAtlas establishes a foundation for real-time dependency analysis.

---

### InfoAtlas: A Foundation Model for Zero-Shot Statistical Dependence Estimate

Authors: Zhengyang Hu, Yanzhi Chen, Hanxiang Ren, Qunsong Zeng, Youyi Zheng, Adrian Weller, Kaibin Huang, Yanchao Yang
Published: 2026-05-29
Categories: cs.LG, cs.AI, stat.ML
arXiv: https://arxiv.org/abs/2606.00241v1

Measuring statistical dependency between high-dimensional random variables is a fundamental task in data science and machine learning. Neural mutual information (MI) estimators offer a promising avenue, but they typically require costly iterative optimization for each new dataset, making them impractical for real-time applications. We present InfoAtlas, a foundation model-like architecture that eliminates this bottleneck by directly inferring MI in a single forward pass. Pretrained on large-scale synthetic data with rich dependence patterns, InfoAtlas learns to identify diverse dependence structures and predict MI directly from the dataset. Comprehensive experiments demonstrate that InfoAtlas matches state-of-the-art neural estimators in accuracy while achieving $100\times$ speedup, can flexibly handle varying dimensions and sample sizes through a single unified model, and generalizes effectively to complex, real-world scenarios. By reformulating MI estimation as an inference task, InfoAtlas establishes a foundation for real-time dependency analysis.

---

### MindZero: Learning Online Mental Reasoning With Zero Annotations

Authors: Shunchi Zhang, Jin Lu, Chuanyang Jin, Yichao Zhou, Zhining Zhang, Tianmin Shu
Published: 2026-05-29
Categories: cs.AI, cs.MA
arXiv: https://arxiv.org/abs/2606.00240v1

Effective real-world assistance requires AI agents with robust Theory of Mind (ToM): inferring human mental states from their behavior. Despite recent advances, several key challenges remain, including (1) online inference with robust uncertainty updates over multiple hypotheses; (2) efficient reasoning suitable for real-time assistance; and (3) the lack of ground-truth mental state annotations in real-world domains. We address these challenges by introducing MindZero, a self-supervised reinforcement learning framework that trains multimodal large language models (MLLMs) for efficient and robust online mental reasoning. During training, the model is rewarded for generating mental state hypotheses that maximize the likelihood of observed actions estimated by a planner, similar to model-based ToM reasoning. This method thus eliminates the need for explicit mental state annotations. After training, MindZero internalizes model-based reasoning into fast single-pass inference. We evaluate MindZero against baselines across challenging mental reasoning and AI assistance tasks in gridworld and household domains. We found that LLMs alone are insufficient; model-based methods improve accuracy but are slow, costly, and limited by backbone MLLM capacity. In contrast, MindZero enhances MLLMs' intrinsic ToM ability and significantly outperforms model-based methods in both accuracy and efficiency, showing that mental reasoning can be effectively learned as a self-supervised skill.

---

### Civilizational Metamaterials: Engineering Coordination Under Capability Gradients and Structural Turbulence

Authors: David Orban
Published: 2026-05-29
Categories: physics.soc-ph, cs.AI, cs.CY, cs.MA
arXiv: https://arxiv.org/abs/2606.00235v1

We argue that governance must transition from a normative discipline to an engineering discipline, and we develop a formal framework, inspired by the physics of metamaterials, to make this transition quantitative and testable. Artificial General Intelligence affects civilization primarily by increasing decision velocity while human verification capacity remains bounded. When the cost of validating AI-generated outputs exceeds the expected utility of acting on them, rational agents default to inaction: a stable but catastrophic Nash equilibrium we term the Freezing Equilibrium. Drawing on metamaterials, where emergent macro-properties arise from designed microstructure, we develop a phenomenological constitutive law for institutional coordination: $R_{\mathrm{eff}} = β\cdot (1-ρ) \cdot (1-τ) \cdot (1-γρτ)$, where $β$ is the decision branching factor, $ρ$ is provenance fidelity, $τ$ is the verification rate, and $γ\in (0,1)$ captures correlated-detection synergy between provenance and verification failures. The model predicts a sharp phase transition between self-healing ($R_{\mathrm{eff}} < 1$) and self-destabilizing ($R_{\mathrm{eff}} > 1$) regimes. We introduce a three-class provenance taxonomy: cryptographic, institutional, and context binding, and derive four falsifiable hypotheses with a proposed 12-week stepped-wedge cluster-randomized trial in government grant review panels. The framework bridges AI alignment theory and institutional design.

---

### Self-Correction Can Amplify Hallucinations: Fact-Level Repair with Graph-Based Evidence Routing in Multimodal Generation

Authors: Kaixiang Zhao, Tianrun Yu, Shawn Huang, Porter Jenkins, Yushun Dong, Amanda Hughes
Published: 2026-05-29
Categories: cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.00232v2

We study fact-level repair for multimodal generation, where a fluent output may contain specific facts that are not supported by the input. Existing inference-time repair methods often generate feedback by jointly conditioning on the input and the current output. This design has two limitations: hallucinated claims in the output can bias the model's interpretation of the input, and free-form feedback cannot be ranked or scheduled at the fact level. We present TIGER, an inference-time framework that redesigns feedback for localized repair. TIGER independently extracts an observation graph from the input and a claim graph from the current output, then assigns each claim a graph-conditioned risk score based on support and conflict. The model repairs selected high-risk claims while keeping the backbone frozen. We provide a convergence analysis showing that the expected total risk decreases geometrically to an explicit asymptotic bound under mild assumptions. Experiments across four cross-modal paths, including image-to-text, image+text-to-text, audio-to-text, and video-to-text, show that TIGER reduces unsupported content while preserving task quality. The gains hold across multiple backbones, and a CrisisFACTS case study suggests that the same repair mechanism can improve grounding in multi-source settings.

---

### TIGER: Traceable Inference with Graph-Based Evidence Routing for Mitigating Hallucinations in Multimodal Generation

Authors: Kaixiang Zhao, Tianrun Yu, Shawn Huang, Porter Jenkins, Yushun Dong, Amanda Hughes
Published: 2026-05-29
Categories: cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.00232v1

We study fact-level repair for multimodal generation, where a fluent output may contain specific facts that are not supported by the input. Existing inference-time repair methods often generate feedback by jointly conditioning on the input and the current output. This design has two limitations: hallucinated claims in the output can bias the model's interpretation of the input, and free-form feedback cannot be ranked or scheduled at the fact level. We present TIGER, an inference-time framework that redesigns feedback for localized repair. TIGER independently extracts an observation graph from the input and a claim graph from the current output, then assigns each claim a graph-conditioned risk score based on support and conflict. The model repairs selected high-risk claims while keeping the backbone frozen. We provide a convergence analysis showing that the expected total risk decreases geometrically to an explicit asymptotic bound under mild assumptions. Experiments across four cross-modal paths, including image-to-text, image+text-to-text, audio-to-text, and video-to-text, show that TIGER reduces unsupported content while preserving task quality. The gains hold across multiple backbones, and a CrisisFACTS case study suggests that the same repair mechanism can improve grounding in multi-source settings.

---

### A Pre-Training Analogue of Grokking in Language Models: Tracing Delayed Grammatical Generalization

Authors: Sherin Muckatira, Namrata Shivagunde, Vijeta Deshpande, Anna Rumshisky
Published: 2026-05-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.00230v1

Grokking, the phenomenon in which neural networks generalize long after fitting their training data, has been studied in supervised settings on many epochs. LLM pre-training instead involves next-token prediction over an unlabeled corpus, with limited data repetition and no explicit train/validation split. To address this, we propose an exposure-based framework that enables the study of grokking-like dynamics during LLM pre-training. We ground our evaluation in BLiMP minimal pairs, which provide controlled grammatical contrasts. For every BLiMP minimal pair, we identify a critical phrase, the smallest continuous span that captures the grammatical contrast and the phenomenon-relevant context. Examples whose critical phrase appears in the pre-training window are assigned to the proxy-train split; the remaining examples are assigned to the proxy-validation split. Across five grammatical phenomena, we observe delayed generalization. Analyzing pre-training checkpoints before and after generalization shows that grammatical concept vectors become more predictive of grammatical acceptability and occupy a higher-dimensional subspace after generalization. We also find that attention from the critical token to the relevant context token is concentrated in a small number of heads.

---

### Continuous Reasoning for Vision-Language-Action

Authors: Yueh-Hua Wu, Tatsuya Matsushima, Kei Ota
Published: 2026-05-29
Categories: cs.RO, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.00229v2

Natural language is a powerful reasoning medium for language and vision-language models, but it is mismatched to the granularity of continuous control. Text and explicit subgoals operate at task-level granularity, whereas vision-language-action (VLA) policies must choose actions at a much finer temporal scale; a single reasoning step can therefore span many action chunks while remaining only weakly coupled to the action needed now. This suggests a different question for VLA: what should play the role of language? We argue that a useful VLA reasoning medium must be shareable across model instances, verifiable through downstream action improvement, and aligned with temporally extended control structure. Based on this view, we propose Continuous Reasoning for Vision-Language-Action. Our model first predicts continuous reasoning in the form of a structured set of continuous thoughts, then reuses them as shared context for chunk-structured action generation. Better action prediction alone does not certify good reasoning: if the same internal medium cannot be shared across model instances and independently verified through improved downstream control, the added latent may simply become a model-private shortcut that helps on seen behaviors without supporting generalizable control. We therefore instantiate continuous reasoning as a shared Gaussian latent interface and train it with a self-verification objective in which an exponential-moving-average teacher must successfully consume the student's reasoning when predicting target actions. Empirically, Continuous Reasoning improves LIBERO-PRO robustness and performs strongly on real robots, raising mean subtask success over π0.5 by 40.4% on TX-G2, an AgiBot G2-compatible variant, and 26.3% on HSR. This suggests that reasoning in VLA is less about extra tokens than about a shareable, verifiable internal language for action.

---

### LithoGRPO: Fast Inverse Lithography via GRPO Reinforced Flow Matching

Authors: Yao Lai, Xuyuan Xiong, Zeyue Xue, Guojin Chen, Jing Wang, Xihui Liu, Rui Zhang, Robert Mullins, Bei Yu, Ping Luo
Published: 2026-05-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.00228v1

In semiconductor manufacturing, lithography projects circuit layouts onto silicon wafers through an optical mask. As circuit features shrink below the wavelength of light, optical diffraction causes the printed patterns to deviate from their intended layouts. Inverse Lithography Technology (ILT) addresses this challenge by generating optimized masks that enhance the fidelity of pattern transfer onto wafers. While ILT resembles an image synthesis task, its reliance on explicit physical metrics for mask evaluation limits the applicability of existing generative models. We introduce LithoGRPO, an ILT framework that integrates the flow-matching paradigm with GRPO-based reinforcement learning (RL) fine-tuning, enabling efficient exploration of diverse masks for a given target layout. Unlike purely generative or optimization-based approaches, RL in LithoGRPO exploits the explicitly defined, physics-based reward function of ILT, enabling optimization under complex, process-aware constraints. To the best of our knowledge, this is the first framework that unifies flow matching and RL for mask optimization. To improve RL sampling efficiency, we propose a fast shot-counting algorithm for manufacturability evaluation, achieving over 130x speedup while preserving the mask ranking of the traditional shot-count metric. Extensive experiments demonstrate that LithoGRPO achieves state-of-the-art performance over both optimization-based and learning-based methods, while maintaining efficient mask generation.

---

### SEMBridge: Tagless-Final Program Semantics with Weakest-Precondition and Bounded-Checking Interpretations

Authors: Eric Liang
Published: 2026-05-29
Categories: cs.PL, cs.AI
arXiv: https://arxiv.org/abs/2606.00220v1

Formal methods provide rigorous accounts of program behavior, but practical software engineering often works through executable libraries, tests, and incremental design. This paper presents SEMBridge, a small tagless-final framework for generating weakest-precondition and bounded-checking interpretations from the same executable object programs. Instead of committing a program semantics to one abstract syntax tree and then writing separate traversals, object programs are written once against a semantic interface and interpreted into multiple meanings: readable code, concrete execution, predicate transformers, bounded counterexample search, and future proof-assistant or SMT back ends. The Python prototype implements a loop-free imperative core with assignments, conditionals, assumptions, and assertions. Across five example programs, the same tagless-final definitions generated executable state transformers and verification conditions that passed bounded checking over domains up to 729 states. The contribution is not a Scala code-generation system or a new verifier, but a compact architecture for keeping executable semantics, weakest-precondition artifacts, and bounded validation synchronized.

---

### 21cmEMUv3: a hybrid diffusion-LSTM emulator of 21cmFAST summary observables

Authors: Daniela Breitman, Andrei Mesinger, Steven G. Murray, Ivan Nikolic, Roberto Trotta
Published: 2026-05-29
Categories: astro-ph.CO, astro-ph.GA, cs.LG
arXiv: https://arxiv.org/abs/2606.00219v1

We are witnessing a surge in observations of the cosmic dawn (CD) and epoch of reionisation (EoR), driving an increasing demand for fast and robust theoretical interpretation frameworks. In response, machine learning (ML), and emulation in particular, has emerged as a powerful approach to accelerate and enhance inference pipelines. In this work, we present 21cmEMUv3, an emulator trained on 21cmFASTv3 simulations that model both atomically and molecularly cooling galaxies. 21cmEMUv3 is conditioned on $σ_8$ and ten astrophysical parameters to produce seven summary observables: (i) the cylindrical 21cm power spectrum (PS), emulated for the first time at such high resolution and accuracy across a wide redshift range of $z \sim$ 6--30; (ii) the spherically-averaged 21cm PS; (iii) the mean neutral fraction of the intergalactic medium (IGM); (iv) the mean 21cm spin temperature; (v) the global 21cm signal; (vi) the ultraviolet (UV) luminosity functions (LFs); and (vii) the Thomson scattering optical depth. Notably, the cylindrical 21cm PS is emulated via score-based diffusion, while the remaining six summaries are emulated via long-short term memory (LSTM) networks, all achieving sub-percent median accuracy. We use the emulator to reinterpret current 21cm PS upper limits from HERA, for the first time using state-of-the-art hydrodynamical simulations to inform priors on star formation inside molecularly cooling galaxies. We find that our inferred soft-band X-ray luminosity per unit star formation rate is consistent with extrapolations of high-mass X-ray binaries to the low-metallicity regimes expected in the first galaxies, excluding values below $10^{39.2}$ erg s$^{-1}M^{-1}_\odot \rm{yr}$ at $95\%$ confidence. Finally, we produce forecasts for the detection of the cosmic 21cm PS with the Square Kilometre Array for different array configurations. The 21cmEMU package is publicly available.

---

### Quantized Reasoning Models Think They Need to Think Longer, but They Do Not

Authors: Sanae Lotfi, Polina Kirichenko, Steven Li, Zechun Liu
Published: 2026-05-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.00206v1

Post-training quantization (PTQ) is widely used to deploy large language models efficiently, but its effect on reasoning models is not well understood. Across math, coding, and science QA, we find that aggressive PTQ reduces accuracy while increasing chain-of-thought (CoT) length. Surprisingly, we show that in up to 52% of the quantized models' failures, models reach the right answer in intermediate reasoning steps but do not output it as a final answer. To understand why quantization leads to this increase in overthinking errors, we measure the token-level KL divergence between quantized and full-precision output distributions. Positions with high KL divergence correlate strongly with high next-token entropy, and at these positions quantized models disproportionately sample overthinking markers such as "wait", "but", and "alternatively". We show that simply introducing a training-free logit penalty on a curated set of overthinking markers can reduce CoT length by 12--23% while preserving or improving accuracy across 5 models (1.5B-32B parameters), 3 quantization methods, and 5 benchmarks, yielding a favorable Pareto frontier of accuracy against reasoning cost compared to penalizing other token sets. Overthinking errors produced by quantized models are particularly reduced by up to 58%.

---

### APE: Agentic Prompt Enhancer for Image Generation and Editing

Authors: Zijian Huang, Jay Zhangjie Wu, Zian Wang, Tianshi Cao, Jiasi Chen, Sanja Fidler, Huan Ling, Xuanchi Ren
Published: 2026-05-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2606.00204v1

Natural language has become a powerful interface for image generation and editing, yet text-guided visual systems remain highly sensitive to prompt formulation. Semantically similar requests can produce different outputs depending on wording, specificity, and how explicitly visual constraints are stated, motivating prompt enhancement as a trainable component rather than a peripheral user choice. Existing strong enhancers often rely on large, proprietary LLMs such as ChatGPT or Gemini, adding cost, latency, and deployment dependence to the visual generation pipeline. We propose Agentic Prompt Enhancer (APE), a lightweight framework that post-trains small language models (SLMs) as prompt-enhancement agents. APE supports both single-agent rewriting and role-specialized multi-agent enhancement. Its single-agent instantiation, SAPE, rewrites the prompt in one pass, while its multi-agent instantiation, MAPE, decomposes enhancement into a router--rewriter--composer process for handling compositional constraints over objects, attributes, spatial relations, and edits. With task-aware rewards and post-training protocols, APE improves visual alignment and prompt following without modifying the downstream visual model. Experiments on challenging image generation and editing benchmarks demonstrate that post-trained small prompt enhancers reliably outperform their base counterparts, narrowing the gap to closed-source prompt enhancers; in addition, MAPE proves particularly strong on complex compositional tasks within these benchmarks.

---

### DeSQ: Decomposition-based SPARQL Query Generation

Authors: Papa Abdou Karim Karou Diallo, Aditya Sharma, Neshat Elhami Fard, Amal Zouaq
Published: 2026-05-29
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.00203v1

Dominant approaches to Knowledge Base Question Answering (KBQA) fall into two categories. First is the generation of a formal query that suffers from brittleness and limited explainability, and the second is direct answer retrieval through KB exploration that is computationally costly and prone to hallucination. To combine the strengths of both paradigms while mitigating their respective weaknesses, we introduce DeSQ (Decomposition-based SPARQL Query Generation), a KB-agnostic framework that operates in three stages. First, it decomposes complex questions into Atomic Constraints (ACs) that mirror the relational structure of the underlying KB. Second, it generates a two-part structured output: (a) Mapping of each AC to its corresponding SPARQL Fragment, using standardized variable and URIs placeholders, and (b) URIs Grounding block describing each placeholder. Third, it assembles these fragments into a complete SPARQL query. DeSQ surpasses state-of-the-art approaches on four out of five major benchmarks and demonstrates superior robustness to lexical variation. Beyond performance gains, our framework greatly simplifies evaluation by eliminating the need for a live KB endpoint, and its structured output enables fine-grained error analysis, allowing more targeted interventions for improvement.

---