Explore the Frontier of Global Academia

01.

medRxiv (Medicine) 2026-06-24 DOI: HASH:ca74d7d1637a1537db60046f59ec3d84

Allostatic load modifies neuropsychiatric risk following traumatic brain injury

Authors:

Wroblewski↗T. H↗Barr↗P. B↗Bigdeli↗T. B↗Barthelemy↗E. J↗

Importance: Outcomes following traumatic brain injury (TBI) vary substantially, with a subset of individuals experiencing neuropsychiatric morbidity and worse prognosis. Exposure to psychosocial and environmental stressors may be an important, yet understudied, modifier of TBI trajectory. Allostatic load (AL) represents the cumulative physiological burden of chronic stress and provides a useful framework for evaluating pre-injury vulnerability. Objective: To assess the relationship between pre-injury AL burden and risk of mortality and incident neuropsychiatric diagnosis following TBI. Design, Setting, and Participants: This cohort study leveraged electronic health record, survey, and laboratory data from the All of Us Research Program, version 8. Participants aged 18 years or older enrolled between May 6, 2018, and October 1, 2023, were queried for TBI diagnosis using clinical diagnostic codes. Data were analyzed between November 11, 2024, and January 7, 2026. Exposure: The physiological burden of pre-injury chronic stress exposure was estimated using an AL index (pALI) derived from anthropometric and laboratory biomarkers collected before index TBI. Main Outcomes and Measures: Post-TBI mortality and incident neuropsychiatric diagnosis clusters. Mortality risk was assessed using Cox proportional hazards models (hazard ratio [HR] with 95% CI), and risk of incident neuropsychiatric diagnosis was modeled using competing-risk regression with death as a competing event (sub-distribution HR with 95% CI). Results: The primary cohort included 4,552 individuals with an established TBI diagnosis and sufficient biomarker data to estimate pALI. The pALI measure differed across sociodemographic groups and was positively correlated with perceived stress (r=.08, p=.002). Higher pALI was associated with increased post-TBI mortality risk (adjusted HR=1.71; 95%CI, 1.36-2.14). Elevated pALI was also associated with greater risk of incident post-traumatic stress disorder (PTSD; adjusted HR=1.28; 95%CI, 1.10-1.50) and sleep disorder (adjusted HR=1.42 95%CI, 1.29-1.57) diagnoses. Conclusions and Relevance: Higher pre-injury ALI was associated with increased risk of mortality and select neuropsychiatric outcomes following TBI, suggesting that AL burden may shape post-injury trajectories. Pre-injury chronic stress exposure and underlying stress biology may represent underrecognized determinants of vulnerability and resilience in brain injury recovery.

Read & Discuss → View Source →

02.

arXiv (CS.LG) 2026-06-17 DOI: arXiv:2602.04901

Beyond Independent Genes: Learning Module-Inductive Representations for Single-Cell Gene Perturbation Prediction

Authors:

Jiafa Ruan↗Ruijie Quan↗Liyang Xu↗Zongxin Yang↗Yi Yang↗

arXiv:2602.04901v2 Announce Type: replace-cross Abstract: Predicting transcriptional responses to genetic perturbations is a central problem in functional genomics. In practice, perturbation responses are rarely gene-independent but instead manifest as coordinated, program-level transcriptional changes among functionally related genes. However, most existing methods do not explicitly model such coordination, due to gene-wise modeling paradigms and reliance on static biological priors that cannot capture dynamic program reorganization. To address these limitations, we propose scBIG, a module-inductive perturbation prediction framework that explicitly models coordinated gene programs. scBIG induces coherent gene programs from data via Gene-Relation Clustering, captures inter-program interactions through a Gene-Cluster-Aware Encoder, and preserves modular coordination using structure-aware alignment objectives. These structured representations are then modeled using conditional flow matching to enable flexible and generalizable perturbation prediction. Extensive experiments on multiple single-cell perturbation benchmarks show that scBIG consistently outperforms state-of-the-art methods, particularly on unseen and combinatorial perturbation settings, achieving an average improvement of 6.7% over the strongest baselines. The code is available at https://github.com/ttruan2426-dot/scBIG.

Read & Discuss → View Source →

03.

arXiv (CS.AI) 2026-06-24 DOI: arXiv:2606.24745

Beyond U-Net: A Latent-Representation-Aligned Skip-Free Backbone for Flow-Matching Speech Enhancement

Authors:

Wangyi Pu↗Michele Scarpiniti↗

arXiv:2606.24745v1 Announce Type: cross Abstract: Generative models, particularly diffusion and score-based approaches, have recently achieved strong performance in speech enhancement, but their iterative sampling process limits real-time deployment. Flow Matching offers an efficient alternative by transporting noisy speech toward clean speech through an ordinary differential equation with few function evaluations. In this work, we propose a skip-free encoder-decoder backbone for flow-matching speech enhancement, guided by Latent Representation Alignment (LRA). Instead of relying on U-Net skip connections, which may transfer noise-correlated low-level features to the decoder, the proposed model aligns its bottleneck and decoder representations with clean latent features extracted from a frozen Descript Audio Codec encoder-decoder without quantization. This codec-aligned supervision promotes compact clean-speech representations while preserving efficient few-step inference. Experiments on WSJ0-CHiME3 and VoiceBank-DEMAND show improved PESQ and perceptual quality, especially on VoiceBank-DEMAND, using only five function evaluations.

Read & Discuss → View Source →

04.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2503.06637

CLAD: Constrained Latent Action Diffusion for Vision-Language Procedure Planning

Authors:

Lei Shi↗Andreas Bulling↗

We propose CLAD, a Constrained Latent Action Diffusion model for vision-language procedure planning in instructional videos. Procedure planning is the challenging task of predicting intermediate actions given a visual observation of a start and a goal state. However, future interactive AI systems must also be able to plan procedures using multi-modal input, e.g., where visual observations are augmented with language descriptions. To tackle this vision-language procedure planning task, our method uses a Variational Autoencoder (VAE) to learn the latent representation of actions and observations as constraints and integrate them into the diffusion process. This approach exploits that the latent space of diffusion models already has semantics that can be used. We use the latent constraints to steer the diffusion model to better generate actions. We report extensive experiments on the popular CrossTask, Coin, and NIV datasets and show that our method outperforms state-of-the-art methods by a large margin. By evaluating ablated versions of our method, we further show that the proposed integration of the action and observation representations learnt in the VAE latent space is key to these performance improvements.

Read & Discuss → View Source →

05.

arXiv (CS.LG) 2026-06-24 DOI: arXiv:2606.09313

Machine-Learning Emulation of Satellite Greenhouse Gas Retrievals: Stability over Time

Authors:

Nugzar Gognadze↗Motonobu Kanagawa↗Yu Someya↗Hisashi Yashiro↗

arXiv:2606.09313v2 Announce Type: replace Abstract: Retrieval algorithms are used to estimate atmospheric concentrations of greenhouse gases (GHGs), such as carbon dioxide (CO2) and methane (CH4), by solving inverse problems from high-spectral-resolution satellite radiance measurements. However, these algorithms are computationally expensive, which makes real-time estimation at scale difficult. Machine-learning models have therefore been proposed as fast emulators of retrieval algorithms. Most existing studies, however, evaluate them only on test data from the same period as the training data. We study the stability over time of such emulators using data from the Greenhouse Gases Observing SATellite (GOSAT). We show that prediction accuracy generally deteriorates when the test period moves away from the training period. We also show that including time as an input feature substantially improves XCH4 prediction for Lasso and neural-network models. Among the methods considered, a simple Lasso model performs as well as or better than more complex methods such as neural networks, and yields more stable predictions over time. We further validate the results using the Total Carbon Column Observing Network (TCCON), a ground-based observation network. On the TCCON-matched dataset, the time-augmented Lasso achieves errors against TCCON that are comparable to the disagreement between GOSAT and TCCON for both XCO2 and XCH4.

Read & Discuss → View Source →

06.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2606.15652

MosaicQuant: Inlier-Outlier Disaggregation for Unified 4-Bit LLM Quantization

Authors:

Yangjia Hu↗Haodong Wang↗Zicong Hong↗Qianli Liu↗Quanxin Shou↗Jian Lin↗Song Guo↗Xiaowei Shen↗Xiangjun Huang↗Dian Wang↗Jian Yang↗

4-bit quantization significantly reduces the memory footprint and accelerates the inference of large language models (LLMs). However, its limited bit-width representation struggles to faithfully capture both dense common values (inliers) and rare large-magnitude values (outliers), causing substantial accuracy degradation. Existing mixed-precision methods mitigate this by retaining outliers in high precision, but at the cost of breaking the uniformity of low-bit execution, introducing precision conversion and extra data movement that undermine practical speedup. We propose MosaicQuant, a unified 4-bit LLM quantization paradigm built on a novel principle of inlier–outlier disaggregation. Rather than elevating outlier precision, MosaicQuant quantizes the full weight matrix into a dense 4-bit base component, where inliers are captured faithfully while outlier are inevitably quantized. A sparse 4-bit residual component is then introduced to compensate for these quantization errors, selectively targeting the most error-critical weight blocks where output distortion is shown to be concentrated. However, a unified representation alone is insufficient, as naïvely executing the sparse residual as a separate kernel still breaks the unified low-bit inference pipeline. To bridge this gap, we introduce ZipperEngine, which fuses sparse block computation into the dense 4-bit GEMM kernel via an overlapped pipeline, unifying not only the representation but also the execution into a single coherent low-bit inference pipeline. Extensive experiments on LLaMA3 and Qwen3 demonstrate that MosaicQuant preserves near-FP16 accuracy while achieving up to $1.24\times$ speedup over the W16A16 baseline.

Read & Discuss → View Source →

07.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2511.05522

AIRMap: AI-Generated Radio Maps for Wireless Digital Twins

Authors:

Ali Saeizadeh↗Miead Tehrani-Moayyed↗Davide Villa↗J. Gordon Beattie Jr↗Pedram Johari↗Stefano Basagni↗Tommaso Melodia↗

arXiv:2511.05522v4 Announce Type: replace-cross Abstract: Accurate, low-latency channel modeling is essential for real-time wireless network simulation and digital-twin applications. Traditional modeling methods like ray tracing are however computationally demanding and unsuited to model dynamic conditions. In this paper, we propose AIRMap, a deep-learning framework for ultra-fast radio-map estimation, along with an automated pipeline for creating the largest radio-map dataset to date. AIRMap uses a single-input U-Net autoencoder that processes only a 2D elevation map of terrain and building heights. Trained on 1.2M Boston-area samples and validated across four distinct urban and rural environments with varying terrain and building density, AIRMap predicts path gain with under 4 dB RMSE in 4 ms per inference on an NVIDIA L40S-over 100x faster than GPU-accelerated ray tracing based radio maps. A lightweight calibration using just 20% of field measurements reduces the median error to approximately 5%, significantly outperforming traditional simulators, which exceed 50% error. Integration into the Colosseum emulator and the Sionna SYS platform demonstrate near-zero error in spectral efficiency and block-error rate compared to measurement-based channels. These findings validate AIRMap's potential for scalable, accurate, and real-time radio map estimation in wireless digital twins.

Read & Discuss → View Source →

08.

Nature (Science) 2026-06-10 DOI: HASH:4fbb7cb6e335b78a92704d7eb1f1509e

Diverse binding poses of agonistic neurotoxins on human Na_v1.6

Authors:

Xiao Fan↗

Voltage-gated sodium (Nav) channels are key targets of various venomous toxins. Deciphering the binding poses and mechanisms of action of representative toxins will help to dissect the functional mechanism of the channels and facilitate therapeutic development targeting Nav channels1,2. Here we present cryo-electron microscopy (cryo-EM) structures of distinct binding poses of three agonistic peptide toxins on the human Nav1.6–β1 channel complex. The globular β-scorpion toxin Cn2 nestles between the extracellular segment of voltage-sensing domain (VSD) in the second repeat of the Nav1.6 core α-unit (VSDII) and the pore extracellular loops in the third repeat of the Nav1.6 core α-unit (ECLIII), where it is stabilized by interactions with both protein regions and the branched N1372-glycan. Cone snail ι-conotoxin RXIA adopts an elongated conformation, spanning VSDI and VSDIV to wrap around the shoulder of the pore domain (PD). The bullet ant-derived toxin δ-paraponeritoxin-Pc1a exists as a transmembrane helix that stands between VSDII and PDIII. Our findings, corroborated by functional characterizations, illustrate the diversity in peptide toxin binding poses and mechanisms of action, link stabilization of the up state of VSDI or VSDII to channel activation, and provide clues to the rational design of selective Nav channel modulators. Structures of the distinct binding poses of three agonistic peptide toxins—bullet-ant-derived toxin δ-paraponeritoxin-Pc1a, cone snail ι-conotoxin RXIA and the globular β-scorpion toxin Cn2—on the human Nav1.6–β1 channel complex illustrate a diversity in binding poses and mechanisms of action.

Read & Discuss → View Source →

09.

arXiv (CS.AI) 2026-06-24 DOI: arXiv:2606.24026

Can Language Model Agents be Helpful Circuit Explainers in Mechanistic Interpretability?

Authors:

Ayan Antik Khan↗Harsh Kohli↗Yuekun Yao↗Huan Sun↗Ziyu Yao↗

arXiv:2606.24026v1 Announce Type: new Abstract: Mechanistic interpretability has made substantial progress in automatically localizing circuits, but explaining what localized components do remains labor-intensive and difficult to standardize. In this work, we study whether language model (LM) agents can assist with this explanation problem once a circuit has already been identified. We introduce AgenticInterpBench, a benchmark for circuit explanation built from 84 semi-synthetic transformer circuits with 163 component-level annotations. We propose HyVE (Hypothesize, Validate, Explain), an agentic explainer that analyzes each component through an iterative loop of observation, hypothesis generation, and causal validation, eventually producing a component-level explanation and a circuit-level task description. Across four LM backbones, HyVE recovers useful component- and task-level explanations, but no backbone is uniformly best. Our analysis shows that strong backbones usually form observation-grounded hypotheses, while failures more often arise later in the validation loop, through incomplete validation plans, code execution errors, or unresolved hypotheses. A case study on an arithmetic circuit in Llama-3-8B shows that the same formulation can extend beyond semi-synthetic benchmarks to naturally trained models. Overall, LM agents are promising circuit explainers, but reliable validation remains the key obstacle.

Read & Discuss → View Source →

10.

arXiv (CS.AI) 2026-06-12 DOI: arXiv:2605.08116

The Safety-Aware Denoiser for Text Diffusion Models

Authors:

Amman Yusuf↗Zhejun Jiang↗Mijung Park↗

arXiv:2605.08116v2 Announce Type: replace-cross Abstract: Recent work on text diffusion models offers a promising alternative to autoregressive generation, but controlling their safety remains underexplored. Existing safety approaches are geared toward autoregressive models and typically rely on post-hoc filtering or inference-time interventions. These are inadequate for effectively addressing safety risks in text diffusion models. We propose the Safety-Aware Denoiser (SAD), a safety-guidance framework in text diffusion models. The SAD modifies the iterative denoising process such that the text sample at the final denoising step is steered toward provably safe regions of the text space. This inference-time method can integrate safety constraints into the denoiser, avoiding computationally expensive retraining of the underlying diffusion model and enabling flexible, lightweight safety guidance. We evaluate the safety of the generated text using the SAD, with respect to hazard taxonomy, memorization, and jailbreak. Experimental results show that SAD substantially reduces unsafe generations while preserving generation quality, diversity, and fluency, outperforming existing methods. These results demonstrate that our safety guidance during denoising provides an effective and scalable mechanism for enforcing safety in text diffusion models.

Read & Discuss → View Source →

11.

arXiv (CS.CV) 2026-06-24 DOI: arXiv:2606.24602

ViTexQA: A Multi-Frame Temporal Perception Dataset for Video Text Question Answering

Authors:

Zhentao Guo↗Chen Duan↗Tongkun Guan↗Zining Wang↗Kai Zhou↗Pengfei Yan↗

Despite remarkable progress in multimodal understanding, current MLLMs still exhibit limitations in video text understanding, particularly when semantics emerge through the integration of temporally distributed textual cues across multiple frames. This perception challenge fundamentally differs from static image text understanding, yet existing datasets fail to capture: the vast majority of questions remain answerable from single frames, inadequately reflecting real-world video text comprehension demands. To address this, we present ViTexQA, a large-scale video-text QA dataset, and FrameThinker for robust multi-frame temporal reasoning. We build ViTexQA via a quality-controlled Chain-of-Thought (CoT) annotation pipeline boosted with temporal constraints; all its QA pairs demand cross-frame text fusion to solve, enforcing true temporal reliance. FrameThinker adopts two-stage training for explicit temporal modeling: CoT-Guided Supervised Fine-Tuning (SFT) generates frame-aware reasoning chains, followed by Temporally-grounded Reinforcement Learning (RL) optimized with multi-frame coherence rewards. Evaluations show our method outperforms SOTA baselines on ViTexQA, lifting ROUGE-L by 6.3%.

Read & Discuss → View Source →

12.

arXiv (CS.LG) 2026-06-18 DOI: arXiv:2606.18963

Online Reward-Punishment Learning from Fixed-Channel Perceptual Event Streams without Environment Rewards

Authors:

Zirong Li↗

arXiv:2606.18963v1 Announce Type: new Abstract: We study online reward-punishment learning when the environment provides no scalar reward or evaluative label. At each step the agent receives only a fixed-channel perceptual packet, and quantities such as pain, energy, contact, damage, or cognitive error are treated as perceptual dimensions whose valence must be inferred from transition consequences. OHIRL separates four roles: M_psi learns next-packet prediction, D_omega models residual dynamics, C_eta is a fixed internal post-transition trajectory evaluator, and B_xi learns to use the resulting value evidence for later policy updates and action scoring. C_eta uses a recovery-positive and persistence/growth-negative residual-regulation orientation; a coefficient-origin audit shows that equal-unit, raw-equal, and random monotone variants preserve more than 92% of the released top-action rankings, while sign inversion preserves 0%. The reward-free protocol exposes observation transitions while withholding environment rewards, delayed external evaluators, success labels, and action-goodness labels. A conditional error decomposition separates B_xi evidence-estimation error from residual policy-optimization error. In a 2x2-XOR packet task, medicine and chili acquire opposite value under visual XOR contexts, and the same pain or spice increase can be positive or negative depending on consequence structure; B_xi reaches 0.952 balanced reward-sign accuracy. In a full online-interleaved audit, M_psi reaches holdout R2=0.907, B_xi reaches 0.940 sign accuracy, and the policy reaches 0.979 optimal-action accuracy, while immediate packet scores, prediction-error rewards, shuffled targets, zero reward, and error-reduction controls collapse. Hidden-reward CartPole and Taxi controls, public-context no-leakage audits, and module-role ablations further test information boundaries and component necessity.

Read & Discuss → View Source →

13.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2604.23628

Characterizing Admissible Objective Functions for Hierarchical Clustering

Authors:

Ryuki Tsukuba↗Kazutoshi Ando↗

arXiv:2604.23628v2 Announce Type: replace-cross Abstract: Hierarchical clustering is a fundamental task in data analysis, but classical methods have long lacked a principled objective function. Dasgupta [STOC~2016] took an important step toward addressing this gap by proposing a well-motivated objective function for cluster trees. Cohen-Addad et al. [J. ACM 2019] subsequently introduced the notion of admissibility: an objective function is admissible if, whenever the input similarity matrix admits generating trees, its minimizers are precisely those generating trees.They also gave a necessary and sufficient condition for admissibility within a family of objective functions based on aggregate intercluster similarity. We refer to this family as sum-type objective functions. However, apart from Dasgupta's original objective function, no explicit admissible objective functions in this family were provided. In this paper, we study admissible objective functions for hierarchical clustering in two directions. For sum-type objective functions, we give a complete characterization when the scaling function is a symmetric polynomial of degree at most two, and we derive sufficient conditions for degree-three polynomials. We also show that the recursive sparsest cut algorithm achieves an O$(\phi)$-approximation ratio for the admissible objective functions covered by our characterization, where $\phi$ is the approximation factor of the sparsest cut subroutine. We then introduce max-type objective functions, where cluster interaction is measured by maximum, rather than aggregate, intercluster similarity. For this class, we characterize which objective functions are admissible for arbitrary symmetric scaling functions and give a complete characterization when the scaling function is a symmetric polynomial of degree at most two.

Read & Discuss → View Source →

14.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2606.11894

Wild3R: Feed-Forward 3D Gaussian Splatting from Unconstrained Sparse Photo Collection

Authors:

Yuto Furutani↗Takashi Otonari↗Kaede Shiohara↗Toshihiko Yamasaki↗

Feed-forward 3D Gaussian Splatting (3DGS) removes the need for time-consuming per-scene optimization required by traditional 3DGS. However, existing feed-forward approaches struggle with real-world photo collections that include diverse lighting conditions and transient objects. In this paper, we present Wild3R, a feed-forward approach for unconstrained sparse photo collections. The main bottleneck is the lack of training data that provides multiple viewpoints, a variety of illuminations, and transient variations necessary for learning robust scene representations. To address this, we introduce the WildCity dataset, which comprises 200 scenes, 170 lighting conditions, and transient objects, resulting in 337,500 images in total. By leveraging the dataset, our model learns appearance consistency across viewpoints conditioned on reference views, while removing transient content. Extensive experiments demonstrate that our method outperforms existing feed-forward approaches and achieves results competitive with prior per-scene optimization-based methods.

Read & Discuss → View Source →

15.

arXiv (math.PR) 2026-06-16 DOI: arXiv:2604.17253

Large Deviations for the Nonlinear Schrödinger Equation with Randomized Quasi-Periodic Initial Data in Higher Dimensions: Subcritical Case

Authors:

Fei Xu↗Yong Li↗

arXiv:2604.17253v2 Announce Type: replace Abstract: We study the cubic weakly nonlinear Schrödinger equation with randomized spatially quasi-periodic initial data in higher dimensions. Under a polynomial decay assumption in Fourier space, we establish a Large Deviations Principle for rogue waves in the so-called subcritical time regime. The proof proceeds in two main steps. We first characterize the distribution of the linear solution and establish the corresponding linear large deviations principle. The lower bound is obtained via pointwise estimates, while the upper bound follows from a combination of truncation and probabilistic arguments. {The method used in this step appears to be new; compare with [GGKS23].} We then perform a detailed combinatorial analysis of the Picard iteration, deriving an effective bound for the Duhamel term and thereby establishing the nonlinear large deviations principle.

Read & Discuss → View Source →

16.

arXiv (CS.LG) 2026-06-19 DOI: arXiv:2509.15822

Phase Transition for Stochastic Block Model with more than $\sqrt{n}$ Communities

Authors:

Alexandra Carpentier↗Christophe Giraud↗Nicolas Verzelen↗

arXiv:2509.15822v3 Announce Type: replace-cross Abstract: Predictions from statistical physics postulate that recovery of the communities in the Stochastic Block Model (SBM) with a fixed number $K$ of communities is possible in polynomial time above, and only above, the Kesten-Stigum (KS) threshold. This conjecture has given rise to a rich literature, proving that non-trivial community recovery is indeed possible in SBM above the KS threshold. Failure of low-degree polynomials (LDP) below the KS threshold was also proven, as long as $K\ll \sqrt{n}$, where $n$ is the number of nodes in the observed graph. When $K\geq \sqrt{n}$, Chin et al.(2025) recently proved that, in a sparse regime, community recovery in polynomial time is possible below the KS threshold by counting non-backtracking paths. This breakthrough led them to postulate a new threshold for the many-communities regime $K\geq \sqrt{n}$. In this work, we provide evidence supporting their conjecture:\\ 1- We prove that, for any graph density, LDP fail to recover communities below the threshold postulated by Chin et al.(2025) ;\\ 2- We prove that community recovery is possible in polynomial time above the postulated threshold, not only in the sparse regime considered in Chin et al.~(2025), but also in moderately sparse regimes, by counting occurrences of some specific motifs inspired by the LDP analysis.\\ In particular, counting self-avoiding paths of length $\log(n)$, which is closely related to spectral algorithms based on the Non-Backtracking operator, is optimal only in the sparse regime. More complex motifs based on the blow-up of a cycle must be considered in denser regimes.

Read & Discuss → View Source →

17.

arXiv (CS.AI) 2026-06-17 DOI: arXiv:2606.17888

MathVis-Fine: Aligning Visual Supervision with Necessity via Progressive Dependency-Guided Training for Multimodal Mathematical Reasoning

Authors:

Wanshi Xu↗Haokun Zhao↗Haidong Yuan↗Songjun Cao↗Long Ma↗

arXiv:2606.17888v1 Announce Type: new Abstract: Chain-of-Thought (CoT) reasoning has extended from purely linguistic domains to multimodal scenarios; however, existing approaches often treat visual inputs as homogeneous or auxiliary signals, failing to capture the intricate and sample-specific dependencies between text and images in mathematical problem-solving. This gives rise to two core issues: first, the supervisory signals for visual content are generalized and coarse-grained, lacking adaptation to the actual necessity of visual information in each sample; second, training feedback becomes inaccurate when visual rewards are uniformly applied without distinguishing the complementary relationships among inputs. These limitations hinder models from achieving precise multimodal reasoning. In this work, we propose a framework for modeling fine-grained visual dependencies in mathematical reasoning. We first construct the MathVis-Fine dataset, augmenting fine-grained visual annotations with visual dependency ratings. Building upon this dataset, we introduce a two-stage progressive visual enhancement training paradigm that balances answer correctness rewards and visual grounding rewards according to the intrinsic visual dependency level of each sample, thereby mitigating reward bias and improving supervision accuracy. Extensive experiments demonstrate that the MathVis-Fine framework effectively enhances visual perception progressively based on visual dependency, offering a more precise training framework for multimodal mathematical reasoning. We will release the dataset upon acceptance.

Read & Discuss → View Source →

18.

arXiv (CS.LG) 2026-06-17 DOI: arXiv:2606.06227

Reward hacking in physical reinforcement learning revealed by turbulent drag reduction

Authors:

Giorgio Maria Cavallazzi↗Miguel P\'erez-Cuadrado↗Alfredo Pinelli↗

arXiv:2606.06227v2 Announce Type: replace-cross Abstract: A reinforcement-learning agent maximises its reward, which can diverge from the outcome its designer intended. In physical control the reward rarely closes that gap, and drag reduction in wall turbulence makes it concrete. A mass-conservation projection couples agents' outputs and erases the per-agent credit the policy gradient needs; a memoryless policy cannot resolve the slow near-wall cycle it acts on; and a pressure-gradient reward pays for nominal drag reduction by pumping power through the wall. Two degenerate controllers achieve large drag reductions while total dissipation rises, so the reported figure can mask a more wasteful flow. We trace each fault to its cause and fix it: a differentiable projection that restores credit, a recurrent policy with a widened sensing stencil, and a reward scored on the true wall power. The corrected controller acts on the flow within a closed energy budget, earning a conservative $17\%$ under honest accounting.

Read & Discuss → View Source →

19.

arXiv (CS.CV) 2026-06-18 DOI: arXiv:2606.19215

GUMP-Net: An interpretable model-data-driven intelligent algorithm for multi-class pelvic segmentation

Authors:

Liheng Wang↗Yinghui Zhang↗Licheng Zhang↗Hailin Xu↗Qiyong Cao↗Chong Chen↗

Pelvic segmentation is one of the most important and fundamental research problems in precise and intelligent diagnosis and treatment, as well as surgical planning and navigation for pelvic fractures. By combining an improved geodesic active contour model with deep neural networks, we propose GUMP-Net, an interpretable model-data-driven intelligent algorithm for multi-class pelvic segmentation, in which three network modules are designed to constitute the overall segmentation framework together: the object detection module for automatic level set initialization, the edge detector module for learning an anatomy-aware edge detector function and the iteration module for deep level set evolution. Leveraging the advantages of level set representation and deep learning, GUMP-Net shows more accurate, robust and consistent segmentation performance, especially in small training data situation, compared to the state-of-the-art methods. Extensive experiments on pelvic datasets demonstrate the rationality and effectiveness of the proposed algorithm. Further experiments extended to ankle dataset indicate broader applications to other anatomies. The proposed algorithm not only provides an efficient segmentation method for complex fracture reduction, but also gives an interpretable geometric perspective for understanding deep learning segmentation.

Read & Discuss → View Source →

20.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2602.00482

AREAL-DTA: Dynamic Tree Attention for Efficient Reinforcement Learning of Large Language Models

Authors:

Jiarui Zhang↗Yuchen Yang↗Ran Yan↗Zhiyu Mei↗Liyuan Zhang↗Daifeng Li↗Wei Fu↗Jiaxuan Gao↗Shusheng Xu↗Yi Wu↗Binhang Yuan↗

arXiv:2602.00482v2 Announce Type: replace Abstract: Reinforcement learning (RL)-based post-training for large language models (LLMs) is computationally expensive, as it generates many rollout sequences that frequently share long token prefixes. Existing RL frameworks usually process these sequences independently during policy training, i.e., repeatedly recomputing identical prefixes in both the forward and backward passes of policy gradient computation, leading to substantial inefficiencies in computation resources and memory usage. Although prefix sharing naturally induces a tree structure over rollouts, packed tree-mask approaches scale poorly in RL settings. In this paper, we introduce AReaL-DTA, which efficiently exploits prefix sharing in RL training. AReaL-DTA employs a depth-first search (DFS)-based execution strategy that dynamically traverses the rollout prefix tree during both forward and backward computation, materializing only a single root-to-leaf path at a time. To further improve scalability, AReaL-DTA incorporates a load-balanced distributed batching mechanism that dynamically constructs and processes prefix trees across multiple GPUs. On $\tau^2$-bench, AReaL-DTA improves training throughput by up to $8.31\times$ over dense training and up to $1.70\times$ over sparse training. Our code is available at https://github.com/areal-project/AReaL/tree/feat/dta.

Read & Discuss → View Source →

21.

arXiv (CS.CL) 2026-06-15 DOI: arXiv:2512.22671

Fragile Knowledge, Robust Instruction-Following: The Width Pruning Dichotomy in Llama-3.2

Authors:

Pere Martra↗

Structured width pruning of GLU-MLP layers in Llama-3.2 models, guided by the Peak-to-Peak Magnitude (PPM) criterion, reveals a systematic dichotomy in how reducing the expansion ratio affects different model capabilities. While performance on tasks relying on parametric knowledge (e.g., MMLU, GSM8K) and perplexity metrics degrades predictably with decreasing expansion ratios, instruction-following capabilities improve at the 2.4x equilibrium ratio (IFEval: +4.8 points / +46% in Llama-3.2-1B and +3.7 points / +39% in Llama-3.2-3B), and multi-step reasoning remains robust (MUSR). This pattern, observed consistently across both evaluated model sizes, challenges the prevailing assumption in compression research that pruning induces uniform degradation. To investigate this, we evaluated seven expansion ratio configurations using comprehensive benchmark suites that assess factual knowledge, mathematical reasoning, language comprehension, instruction-following, and truthfulness. Our analysis identifies the expansion ratio as a critical architectural parameter that selectively reshapes the model's task performance profile, rather than merely serving as a compression metric.

Read & Discuss → View Source →

22.

arXiv (CS.LG) 2026-06-17 DOI: arXiv:2602.23116

Provably Efficient Regularized Online RLHF with Generalized Bilinear Preferences

Authors:

Junghyun Lee↗Minju Hong↗Kwang-Sung Jun↗Chulhee Yun↗Se-Young Yun↗

arXiv:2602.23116v3 Announce Type: replace Abstract: We consider the problem of regularized best-response max-regret minimization in online RLHF under general preferences and bandit feedback. While various regularizers are utilized to robustify alignment, known polylogarithmic regret guarantees remain heavily specific to KL. To investigate whether such fast rates extend beyond KL, we adopt the Generalized Bilinear Preference Model (GBPM) – capturing intransitive preferences over $d$-dimensional item-wise features via a rank-$2r$ skew-symmetric matrix – to isolate the impact of generic regularization. Crucially, under GBPM, we prove that the dual gap of any greedy policy is bounded by the squared estimation error, derived using only strong convexity and skew-symmetry. Under a feature coverage assumption, we establish a generic polylogarithmic regret of $\tilde{\mathcal{O}}(\eta d^4 C_{\min}^{-1} (\log T)^2 \wedge d^2 C_{\min}^{-1/2} \sqrt{T})$ with Greedy Sampling, and a dimension-wise improved regret (for well-conditioned arm-sets) of $\tilde{\mathcal{O}}(C_{\min}^{-2} \sqrt{\eta r T} \wedge r^{1/3} C_{\min}^{-4/3} T^{2/3})$ with Explore-Then-Commit, where $\eta^{-1}$ is the regularization coefficient, $T$ is the time horizon, and $C_{\min}$ is an arm-set dependent quantity. This demonstrates that ``fast'' regrets are not KL-specific, but rather a fundamental consequence of generic strongly convex geometry.

Read & Discuss → View Source →

23.

arXiv (CS.AI) 2026-06-11 DOI: arXiv:2606.10794

READER: Robust Evidence-based Authorship Decoding via Extracted Representations

Authors:

Jiaxu Liu↗Sunnan Mu↗Dong Huang↗Liuyin Wang↗Jing Shao↗Jie Zhang↗

arXiv:2606.10794v2 Announce Type: replace Abstract: As agentic applications increasingly route user tasks through official and third-party LLM APIs, provenance becomes an operational question: which model generated a given black-box response? We study Dynamic Black-Box LLM Provenance: identifying the source LLM from generations elicited by query-varying, non-predefined prompts rather than a fixed input set or benchmark suite. This setting is difficult because prompt semantics dominate the text, while model-specific authorship traces are weak and inconsistent at the surface level. We introduce READER (Robust Evidence-based Authorship Decoding via Extracted Representations), a lightweight provenance framework that treats a frozen proxy LLM as a reader of hidden authorship evidence. READER maps black-box outputs into proxy activation space, temporally filters token states within each response, and performs Bayesian Evidence Accumulation by summing single-response log-posterior evidence across independently sampled prompts. This avoids fragile mean-pooling of prompt-specific representations while preserving the query-wise evidence needed for calibrated confidence. On Agent500, a 50-target dataset built from agent-style prompts, READER reaches $31.0$-$42.4\%$ top-1 accuracy from a single response and $70.0$-$84.0\%$ from 50 responses, substantially outperforming sentence-encoder fingerprints. Scaling across nine proxy readers further shows that stronger LLMs expose more linearly decodable authorship structure, suggesting that authorship perception is already present in frozen LLM representations and can be converted into reliable multi-query attribution.

Read & Discuss → View Source →

24.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.15992

Multi-Task Tennis Stroke Biomechanics Analysis Using MediaPipe Pose

Authors:

Jigyashman Hazarika↗

We built a multi-task pipeline for tennis stroke biomechanics from plain RGB video. On top of pose-based stroke recognition, it adds two new tasks, predicting shot direction and grading posture quality, plus a rule-based feedback layer that suggests coaching tips. Strokes are found automatically using a weighted joint velocity score, s(t) = 0.5 v_wrist + 0.3 m_elbow + 0.2 m_shoulder, removing the need for manual annotation. Pose comes from MediaPipe Pose Landmarker (33 landmarks, metric world coordinates), with each stroke turned into a 30-frame by 39-feature sequence for TennisTransformerGPU, a compact 564,103-parameter transformer (4 layers, 4 heads, d=128) with three parallel output heads. Trained on 1,281 labeled strokes from 7 pros and 1 amateur across 11 videos, it hits 83.7% stroke-type accuracy, 61.9% on direction, and 62.6% on posture under a random 80/20 split. The interesting test is cross-player: train on pros, evaluate on the amateur. Stroke type barely budges, 82.9%, a 0.8% drop. Direction prediction does not transfer; it just falls back to the majority class. An ablation shows why world coordinates matter so much here: switching to image-space landmarks tanks cross-player stroke-type accuracy from 83% to 47% and direction from 68% to 21%. Everything runs on Kaggle's free T4 GPU tier and is fully reproducible.

Read & Discuss → View Source →

25.

PLOS Computational Biology 2026-06-05 DOI: HASH:d96b2ba76138ac2b74598da82f90086f

A multiscale, Bayesian inference approach to augment mechanistic models of cell signaling with machine-learning predictions of binding affinity

Authors:

Holly A. Huber↗

by Holly A. Huber, Stacey D. Finley Computational models in systems biology are often underdetermined—that is, there is little data relative to the complexity and size of the model. This lack of data is primarily due to limits in our ability to observe specific biological systems and restricts the utility of computational models. To reduce this uncertainty, recent methods have explored augmenting parameter inference of systems biology models with predictions from machine learning models. Such approaches expand the pool of data that is applicable for the inference problem. Here, we explore augmenting the parameter inference of intracellular signaling models. We choose to investigate signaling because experimental measurements of the variables of interest, protein dynamics, are still quite limited. To investigate, we propose a novel, multiscale, Bayesian inference approach that augments traditional signaling data with predictions of binding affinity. These predictions are generated using a machine learning pipeline with measurements of amino acid sequence, from the Universal Protein Resource, or protein structure, from the Protein Data Bank, as inputs. We find that we can successfully integrate these measurements into the inference problem using our novel framework. Excitingly, this integration significantly improves the parameter estimates of signaling models. We demonstrate that how much this improvement impacts predictions of signaling depends on the sensitivity of the prediction to perturbations in the parameter values. Overall, the framework we establish here improves the parameter inference of intracellular signaling models by successfully bridging data on protein sequence and structure with systems-level signaling.

Read & Discuss → View Source →