Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Thursday 1 October

Computer Vision

CtrlWAM: Controllable World Action Models with Aligned Intent and Foresight

Chensheng Peng, Wenhao Ding, Ran Tian et al. (+9 more)

World action models (WAMs) jointly predict actions (intent) and visual future (foresight). Standard training adds noise to recorded actions and video simultaneously, but such training paradigms introduce a mismatch: perturbed actions imply counterfactual future visual, while the noised video remains tied to the GT recording. In low-noise regime, the scene geometry and even the dynamic behavior remain clearly visible from the noisy future frames despite the added noise. We present CtrlWAM, which executes perturbed actions in a simulator and pairs them with their noised visual consequences for joint WAM learning. To accommodate the different denoising requirements of video and actions, we introduce warped video–action noise schedules that aim to keep visual layout responsive as action predictions evolve. We further extend the action interface from ego-only control to a variable number of agent streams, allowing a unified model to represent predicted or commanded futures for multiple agents. Driving experiments show more accurate action forecasts, closer agreement between generated video and actions, and better following of supplied commands; robotics experiments show stronger motion fidelity and controllability. Matched controls support the benefit of off-path renders for command following and manipulation fidelity. Together, these findings contribute to a more controllable world action model. Project page: https://ctrl-wam.github.io/

Read the full abstract

All authors Chensheng Peng, Wenhao Ding, Ran Tian, Zewei Zhou, Jef Packer, Maximilian Igl, Peter Karkus, Yan Wang, Masayoshi Tomizuka, Boris Ivanovic, Marco Pavone, Yuxiao Chen

Machine Learning

Fixing a Model That Learned Worse Cancer Means Lower Risk: Monotonic Constraints in Bladder Cancer Recurrence Prediction

Saram Abbas, David Thomas, Naeem Soomro et al. (+3 more)

Background and Objective: Clinicians expect recurrence risk to climb with cancer severity. In a UK multicentre trial, an unconstrained XGBoost model learnt that higher tumour stage and carcinoma in situ predicted lower recurrence risk, and discrimination, calibration, and SHAP were all blind to it. We developed a counterfactual testing framework to detect this inversion and a monotonic-constraint framework to remove it without hurting performance. Methods: BOXIT enrolled 472 patients with protocol-mandated cystoscopy across 51 UK sites (2007-2012); 435 had at least two years' follow-up (153 recurrences, 35.2%). We developed a counterfactual direction test and a monotonic-constraint correction, with constraint directions drawn from the EORTC and EAU risk systems, and evaluated both against unconstrained XGBoost and logistic regression on 18 predictors (seven directed) over 50 cross-validation folds. The test worsened each patient on one directed feature at a time to check whether risk fell; SHAP direction and calibration were also assessed. Key Findings and Limitations: Tumour stage and carcinoma in situ were associated with lower recurrence, opposite to medical intuition; the unconstrained model reversed carcinoma in situ counterfactuals in 90.2% of cases and stage in 74.3%. Discrimination ($\Delta$AUC 0.005, p=0.47), calibration, and SHAP magnitude were all blind to the inversion. Monotonic constraints eliminated every violation at no cost to discrimination (0.723 vs 0.718) and outperformed EORTC (p=8.9e-16). Limitations: single trial, internal-external validation only. Conclusions and Clinical Implications: A model that had learned this inversion passed every conventional check. A counterfactual direction test, run as a single refit with pre-specified monotonic constraints, catches this failure at no cost to performance and should be routine before clinical deployment.

Read the full abstract

All authors Saram Abbas, David Thomas, Naeem Soomro, Rishad Shafik, Rakesh Heer, Kabita Adhikari

Quantum Gases

Study of quantum turbulence by vortex-antivortex dynamics in dipolar BECs

S. Sabari and Lauro Tomio

We investigate vortex nucleation and dynamics in a dipolar Bose-Einstein condensate stirred by a rotating Gaussian obstacle. Two stirring protocols are compared: continuous stirring with constant amplitude (Case 1), and obstacle removal at $t = 15$ ms with linear amplitude ramp-down (Case 2). In both cases, a smooth angular velocity ramp ($t_{\text{acc}} = 3$ ms) suppresses spurious phonon excitations. The first vortex pairs exit the obstacle at $t \sim 3$ ms, and by $t \sim 8$ - $10$ ms the individual vortices and antivortices are fully resolved in the phase maps. Case 1 yields a vortex population that grows up to $t \sim 40$ ms and then roughly saturates, while Case 2 yields a population that peaks around $t \sim 15$ - $25$ ms and then decreases. Remarkably, both protocols lead to local triangular-like ordering in selected regions of the condensate by $t \sim 50$ ms, exhibiting clear dipolar signatures: elliptical distortion and elongated vortex cores along the polarization axis. The mixed vortex-antivortex population remains stable up to $100$ ms with no observable annihilation, indicating that dipolar interactions strongly suppress vortex decay. Kinetic energy decomposition confirms that incompressible (vortex) energy dominates at late times. These results establish a controlled platform for vortex studies in dipolar superfluids and provide benchmarks for future experiments on quantum turbulence with long-range anisotropic interactions.

Read the full abstract

Number Theory

Transcendence of dendric beta-expansions with pure discrete spectrum

Jihyuk Seo and Wolfgang Steiner

We prove the transcendence of numbers that have $\beta$-expansions (in algebraic bases $\beta$) belonging to dendric shifts with pure discrete spectrum, under combinatorial conditions on their directive sequences of substitutions. This includes Arnoux-Rauzy sequences on arbitrary alphabets and Cassaigne-Selmer sequences. The proof uses Rauzy fractals and the subspace theorem.

Read the full abstract

Computer Vision

Lang3DSeg: Annotation-Free Open-Vocabulary 3D Segmentation with Point Transformers

Cigdem Kokenoz, Amir Salarpour, Alkim Domeke et al. (+5 more)

Accurate 3D semantic perception is critical for safe autonomous navigation. However, supervised LiDAR segmentation remains tied to closed taxonomies and to the cost of point-wise manual annotation. Open-vocabulary methods avoid that cost by projecting the output of 2D vision-language models onto LiDAR and distilling it into a 3D network. These methods rely almost exclusively on voxel-based sparse convolutions, and point transformers have so far been limited to indoor environments, where 3D data is dense and bounded. We present Lang3DSeg, which establishes a point transformer as the backbone for annotation-free open-vocabulary segmentation of outdoor 3D LiDAR, and is trained from scratch without geometric pre-training. This training paradigm necessitates addressing the inherent noise in 2D-to-3D label projections; specifically, naive projection often suffers from depth ambiguity, where points behind an object are erroneously assigned its semantic label. We therefore composite masks using an explicit class-priority rule and truncate each projected instance at the first gap in its depth distribution, correcting the projection error directly rather than averaging it over registered sequences. Lang3DSeg achieves 52.8% mIoU on nuScenes validation and 41.4% on SemanticKITTI, the highest among published annotation-free methods on both benchmarks. Every 3D semantic segmentation is on a single LiDAR sweep, and inference operates in real-time without running vision-language models.

Read the full abstract

All authors Cigdem Kokenoz, Amir Salarpour, Alkim Domeke, Christopher Salas, Pedram MohajerAnsari, Long Cheng, Mert D. Pesé, Bing Li

Robotics

Are Frontier VLM Agents Ready to Be Robot Generalists? An Empirical Study with the Embodied Agent Arena

Haojian Huang, Pukun Zhao, Zexi Li et al. (+7 more)

A robot generalist must interpret a scene, act on it, and satisfy every condition of a task. How far do frontier vision-language models (VLMs) meet these demands? We introduce Embodied Agent Arena, evaluating ten VLM agents across Geometry, Spatial Reasoning, Affordance, Task Planning, and Manipulation. Its 1,000 cases combine 32 established sources with GeoProbe, our 168-case benchmark for camera, motion, depth, and scale estimation. A minimal harness leaves task decisions to the model through source-native interfaces. Closed-weight models lead all five domains, whereas open-weight VLMs show useful spatial reasoning but substantially lower planning and manipulation success. Beneath these averages, similar geometric errors conceal different scale and shape distortions, and valid region estimates need not identify usable action points. During execution, agents can achieve early subgoals yet lose object control, bind an action to the wrong receiving region, or stop before the required physical state holds. Additional computation does not uniformly resolve these failures: code repair can restore execution without completing the task, while self-review can reverse correct judgments. Failure costs also differ: the strongest planners concentrate failed-token expenditure in prolonged body-control tasks, whereas other models fail broadly across procedures. Matched traces show how successful recovery uses feedback to restore a missing prerequisite and advance the remaining task.

Read the full abstract

All authors Haojian Huang, Pukun Zhao, Zexi Li, Yehang Zhang, Yangkai Wei, Wenqian Li, Han Yang, Kaiwen Zhou, Ying-Cong Chen, Yinchuan Li

General Topology

Square-to-Cube Lindelöfness and Power Separations in Hattori Spaces

Xing-Yu Hu

Let $H(A)$ be the Hattori space associated with $A\subseteq\mathbb R$, and put $B=\mathbb R\setminus A$. Kulesza announced at the 2018 Joint Mathematics Meetings that Lindelöfness of $H(A)^{2n}$ forces Lindelöfness of $H(A)^{2n+1}$ for every positive integer $n$. The meeting abstract states this implication without a proof. This paper proves the case $n=1$ in ZFC. The same one-sided geometry gives a pointwise half-space criterion for affine discreteness and, when $H(A)^2$ is Lindelöf, a four-coordinate obstruction for uncountable closed discrete monotone traces. Known results on mixed Sorgenfrey–Euclidean products lead to hereditary-Lindelöf criteria for finite powers, a countable-power characterization, and a finite-power Lindelöf reduction. A transfinite construction gives $A\subseteq\mathbb R$ for which $H(A)^2$ and $H(A)^3$ are Lindelöf, whereas $H(A)^4$ contains a closed discrete affine line of cardinality $\mathfrak c$.

Read the full abstract

Computation and Language

Child-Adapted Structured Phonological Representations for Interpretable Speech Sound Analysis

Abner Hernandez, Tomás Arias Vergara, Andreas Maier et al. (+1 more)

Structured phonological representations provide an interpretable alternative to generic speech embeddings, but existing models are largely trained on adult speech. We adapt PhonoQ-2.0 to child speech using CHILDES-Aligned data and compare three alignment-supervision conditions (Adult, Adult+Child, and Child-only) across two initialization strategies (Adult PhonoQ and scratch). Generalization is evaluated against manual child-speech annotations. On 1,352 consonant targets from 58 typically developing children, child-speech adaptation improves voicing recognition across all supervision conditions, from 0.922 macro-F1 for Adult PhonoQ to 0.972–0.987 after adaptation. Manner is more sensitive to alignment supervision: Adult+Child MFA reaches 0.804 and 0.796, compared to approximately 0.70 under Adult MFA supervision. Place remains comparatively strong across systems (0.871–0.902), although per-class performance varies substantially. The velar-fronting contrast is preserved across all seven model variants. Longitudinal UltraPhonix analysis further reveals speaker-specific velar and post-alveolar changes that are largely preserved across models and broadly consistent with reported clinical progress.

Read the full abstract

All authors Abner Hernandez, Tomás Arias Vergara, Andreas Maier, Paula Andrea Pérez-Toro

Computer Vision

SmoothOperator: Enhancing Representations for Fine-grained Open-set Recognition via Modulated Label Smoothing

Thiru Thillai Nadarasar Bahavan, Yu Xia, Sachith Seneviratne et al. (+1 more)

Open Set Recognition (OSR) aims to enable models to accurately classify known classes while rejecting samples from unseen classes. A key challenge in OSR lies in the inability to model the unbounded distribution of unknown classes during training, often leading to the misclassification of samples from these classes. Rather than modeling unknowns, recent work shapes the feature space so that known classes are compact and well separated, and spherical representation learning methods have achieved strong results this way. Label smoothing has been identified as one of the key drivers of this success, yet it applies the same coefficient to every training sample, regardless of how well each sample is already embedded. We show that the spherical representation learning objectives used in OSR share a single alignment–uniformity structure in which labels enter only through the alignment term. Label smoothing therefore acts as an alignment dial, and a fixed coefficient sets this dial to the same value for every sample. We propose a plug-in, SmoothOperator (SmoothOP), which sets the smoothing coefficient of each sample from its prominence, an embedding-space signal measuring how clearly the sample's own class stands out against its strongest competing class. Our method integrates into four existing spherical representation learning methods at minimal training overhead. SmoothOP assigns strong smoothing to samples with high prominence, which reduces their alignment and relaxes their pull. On the Semantic Shift Benchmark, SmoothOP-augmented variants generally outperform their base objectives across datasets, degrees of semantic shift, and OSR post-processors, with gains of up to 4.7% in AUROC, OSCR, and closed-set accuracy.

Read the full abstract

All authors Thiru Thillai Nadarasar Bahavan, Yu Xia, Sachith Seneviratne, Saman Halgamuge

Cryptography and Security

AuraForge: Scaling Security Supervision for Training Coding Agents

Danqing Wang, Songwen Zhao, Harsh Sharma et al. (+4 more)

Coding agents are now proficient enough to generate complex software applications from a single prompt. As their capabilities have grown, human oversight has increasingly shifted from line-by-line code review toward hands-off evaluation of outcomes. However, recent studies have shown that such a transition exposes a critical risk: functional correctness alone does not guarantee a secure implementation. Despite growing attention to code security, training safer coding agents remains challenging because reliable security supervision is difficult to obtain at scale from real-world repositories. We introduce AuraForge to synthesize and validate executable security tests for training secure coding agents. Our approach combines attack-oriented test synthesis, language-extensible task construction, and safeguards against reward hacking. Using AuraForge, we construct AuraGym, a multi-language and multi-CWE executable training gym: 679 executable feature-implementation tasks from 344 real-world repositories across Python, JavaScript, and TypeScript, covering 177 CWE categories. On the subset with human-written security tests, AuraForge produces about 3 times as many test cases on average and reduces the false-positive rate by 83.23%, allowing alternative secure implementations to receive correct supervision. Training Qwen3.5-4B with synthesized security tests gains larger improvements than human-written security tests (average 19.7 FuncPass and 6.2 SecPass vs. 14.9 FuncPass and 4.4 SecPass) on three languages. These results demonstrate that AuraForge provides more diverse and reliable security supervision to train secure coding agents.

Read the full abstract

All authors Danqing Wang, Songwen Zhao, Harsh Sharma, Jierui Wang, Andre Vicente Duarte, Ivan Bercovich, Lei Li

Artificial Intelligence

Learning Multiple Timescales for Goal-Conditioned Reinforcement Learning

Pedro Robles Dutenhefner, Dikshant Shehmar, Wagner Meira Jr. et al. (+1 more)

Existing approaches to offline goal-conditioned reinforcement learning (GCRL) struggle with long-horizon tasks. Discounting shrinks value differences between distant states until they fall below the function approximation error, leaving the agent with no signal for ranking states. Temporal abstraction, which treats k environment steps as a single transition, restores this signal at long range, but no single fixed k suits all state-goal distances: large k preserves value differences across long temporal distances while collapsing distinctions between nearby states, and small k does the reverse. We make this trade-off explicit and introduce Generalized Implicit Temporal Abstraction (GITA), which conditions a single value function on k. GITA trains one policy by aggregating advantage-weighted supervision across multiple k values, so scales assigning larger positive advantages to a state-goal pair contribute more strongly to its update. GITA does not need to choose between local resolution and long-range signal; it retains both without committing to a single k. On OGBench, GITA outperforms a broad range of offline GCRL baselines, raising average success rate across all tasks by 25 percentage points (73% relative improvement) over HIQL. It also improves over the strongest fixed-k method, OTA, by 7 percentage points (14% relative).

Read the full abstract

All authors Pedro Robles Dutenhefner, Dikshant Shehmar, Wagner Meira Jr., Marlos C. Machado

Computer Vision

Geometric Similarity in VLM Low-Level Vision Representations

Shao-Jun Xia, Huixin Zhang, Zhen Lei et al. (+3 more)

Vision-language models (VLMs) have emerged as powerful candidates for universal vision backbones, with representative architectures including autoregressive (AR) models and diffusion transformers (DiTs). Yet, adapting them efficiently for all-in-one low-level image restoration remains a challenge. Crucially, the field lacks an understanding of how VLMs organize hidden-layer representations and whether these structurally distinct paradigms share a common geometric organization for pixel-level perception. Such shared organization is a prerequisite for building highly transferable, unified restoration VLMs and adapters. In this paper, we systematically investigate representational similarity across 24 low-level tasks spanning 5 categories. We propose GeoSim, a unified four-level framework that analyzes task-conditioned representations from global similarity, local geometry, sparse feature decomposition, and topological verification perspectives. Our formulation applies to the analysis of hidden states in AR models and feature maps in DiTs across same- and cross-task/model settings. Our results reveal the organizing principles of low-level visual representations while exposing their limits in cross-task and cross-model agreement. Ultimately, GeoSim provides an interpretability lens for probing latent transferability in low-level vision and diagnosing model limitations in task- or model-specific scenarios.

Read the full abstract

All authors Shao-Jun Xia, Huixin Zhang, Zhen Lei, Anlan Sun, Yuner Zhang, Xiaoyang Chen

Quantum Physics

Beyond IP = PSPACE and QIP = PSPACE: Interactive Proofs in Arbitrary Physical Theories

Kishor Bharti

The equalities IP = PSPACE and QIP = PSPACE, the latter achievable with three messages, raise a basic question: how much of an interactive proof's power comes from the underlying physical theory? We study interactive proofs in general probabilistic theories, which include classical and quantum theory. The answer depends on what the prover and verifier exchange and how the theory specifies efficient operations. When they exchange only classical messages, protocols in every theory satisfying our standard assumptions decide exactly PSPACE. For protocols with a quantum verifier and quantum messages, allowing a prover to use any theory containing quantum theory does not increase the maximum acceptance probability. Thus, the three-message PSPACE result remains valid against such provers. When messages may be arbitrary systems, the interactive-proof class can strictly exceed PSPACE.

Read the full abstract

Data Structures and Algorithms

Sparsification Framework for Directed Densest Subgraph

Slobodan Mitrović and Theodore Pan

We develop a new approach for computing approximate directed densest subgraphs (DDS). Our main result is a sparsification procedure that reduces a directed graph $G$ on $n$ vertices to a graph with $n \cdot \text{poly} \log n$ edges while preserving enough structure to recover an approximate DDS of $G$. Instantiating this framework in several memory-constrained settings, we obtain the following improvements over the state of the art: In semi-streaming, we obtain a single-pass algorithm that computes a $(1-\varepsilon)$-approximate DDS. Previously, the only semi-streaming algorithm that computed a constant approximation of DDS was by Bahmani, Kumar, and Vassilvitskii (2012), providing a $0.5-\varepsilon$ approximation in $O(\log n)$ passes. Hence, our work completely closes the approximation gap between undirected and directed DS in the semi-streaming setting, matching the $(1-\varepsilon)$-approximate undirected DS algorithm by Esfandiari, Hajiaghayi, and Woodruff (2016). In the near-linear-memory MPC regime, we obtain an $O(1)$-round algorithm for $(1-\varepsilon)$-approximate DDS, improving over the $O(\sqrt{\log n})$-round $(0.5-\varepsilon)$-approximation algorithm of Mitrović and Pan (2024). In the sublinear-time setting, we obtain an algorithm using $\tilde{O}(n)$ time, space, and oracle queries to compute a $(1-\varepsilon)$-approximate DDS, improving over the $\tilde{O}(n^{1.5})$ time, space, and query algorithm of Esfandiari, Hajiaghayi, and Woodruff (2016).

Read the full abstract

Plasma Physics

Is Your AI Fast Enough to Run a Fusion Reactor?

Nathaniel Chen, Cheolsik Byun, Sangkyeun Kim et al. (+6 more)

In magnetic confinement fusion, compact machine learning algorithms are often integrated into the control loop on CPU-based custom hardware stacks. Deterministic implementations of these algorithms target instabilities which can evolve over microseconds to milliseconds. Staying within this framework may therefore limit the pace of control-system development to the inference time available. With the expansion of hardware acceleration and large models, we aim to bring ML models for fusion control to modern inference engines and compare, based on the physics reactor, the suitability of each engine per model. This study implements a consistent real-time machine learning algorithm for GPUs and optimized CPU packages, including OpenVINO, XLA, and TVM, at DIII-D National Fusion Facility. For the smallest model, one CPU core responds within 3.6 $\mu$s in 99% of control cycles and an NVIDIA A100 within 28 $\mu$s. We also evaluate models running simultaneously, as they would run in a typical experiment, and find best use cases per framework.

Read the full abstract

All authors Nathaniel Chen, Cheolsik Byun, Sangkyeun Kim, Andrew Rothstein, Ricardo Shousha, Hiro Josep Farre Kaga, Peter Steiner, Azarakhsh Jalalvand, Egemen Kolemen

High Energy Physics (Theory)

Complexity study of the Hartle-Hawking state in JT gravity

Ritam Basu

The Wigner function, in general, takes on negative values, and the amount of negativity in the Wigner function gives an operationally meaningful measure of the complexity of simulating the quantum state on a classical computer. In this paper, we study the growth of Wigner negativity of the Hartle-Hawking state of Jackiw-Teitelboim gravity under time evolution. We work in the gravitational length basis, at genus zero, and at high temperature, $\beta \ll 1$. Our main analytic result is simple: to leading order in $\beta$ the state is a Gaussian wavepacket of minimum uncertainty, sitting at rest at the turning point of the Liouville wall. Its Wigner function is therefore positive, and the negativity is $1+\delta(\beta,t)$, where the correction $\delta$ is smaller than any power of $\beta$. We show that $\delta$ is an even function of time, so there is no linear growth at $t=0$, and that once the packet has reflected off the wall its leading-order evolution is a Clifford shear, which cannot change the negativity. We also numerically study the time evolution of the Wigner function and its negativity. We observe that $\delta$ stays below $10^{-11}$ at $t=0$ and through the reflection, then rises slowly over a few tens of $\beta/\pi$, and then approaches a late-time value consistent with $\delta_\infty(\beta) \simeq 0.08\, e^{-4.4/\beta}$ for $0.25 \le \beta \le 1$: the lower the temperature, the larger the late-time negativity. The exact survival amplitude $Z(\beta+it)/Z(\beta)$ fixes the spread complexity to order $t^4$, and it also fixes the seed-normalised Wigner diagnostic of our earlier work (arXiv:2607.04065, arXiv:2607.17346). We take this as evidence that the length basis is ideally suited for a dual, semi-classical description of the dynamics at genus zero. Beyond genus zero the length basis is overcomplete, and we make no statement about finite $e^{S_0}$.

Read the full abstract

Wednesday 30 September

Operating Systems

Akira: Towards Operating System Architecture for Agile Specialization

Renshuang Jiang, Pan Dong, Yichong Wang et al. (+6 more)

Operating Systems (OS) specialization tailors abstractions and privileges to specific use cases, enabling configurable tradeoffs between performance and isolation. However, increasing application diversity makes low-cost and non-invasive agile specialization challenging. First, modular OS designs decompose the system into components and reduce coupling, but lack abstractions separating component implementations from OS configuration. Second, diverse workloads require different isolation strengths, yet existing OSs largely fix isolation models at design time, making isolation adaptation changes costly and invasive. To address these, we present Akira, an OS architecture that enables efficiently adapt OS composition and isolation. Akira introduces the Abstraction and Encapsulation Unit (AEU), which separates component interfaces and isolation requirements from implementations, hiding modular OS internals and enabling non-invasive reconfiguration. Based on the abstraction, Akira introduces the Control Logic Unit (CLU) and the Bridge-based Isolation Unit (BIU) to separately manage component communication and isolation, enabling their independent adaptation without system-wide modifications. Results show that Akira enables rapid and low-cost OS reconfiguration with only a 1.10x slowdown over native FreeRTOS and reduces reconfiguration effort from 1.38 KLOC of intrusive changes to 0.28 KLOC of incremental updates. On Unikraft-based Redis and Nginx, Akira configurations expand the performance-isolation space by up to 10.4x and 7.8x. We publicly release Akira's source code, at https://github.com/JiangRS17/Akira.

Read the full abstract

All authors Renshuang Jiang, Pan Dong, Yichong Wang, Lei Li, Yuchen Dong, Xiaoxiang Fang, Qirui Yu, Jianfeng Zhang, Zhe Jiang

Statistics Applications

Separating Weight and Likelihood Tail Effects in Bayesian Neural Groundwater Inversion

Ke-Ming Shen

Groundwater inversion is commonly built on Gaussian assumptions for both network weights and output errors. The assumptions are convenient, but it is unclear which one should be relaxed when hydraulic conductivity is heavy tailed. Here, Gaussian and q-Gaussian distributions were assigned independently to variational weights and the conditional output likelihood in a $2\times2$ Bayesian neural design. Hydraulic conductivity, porosity, and dispersion were inferred jointly from sparse head and transport observations. Four q regimes ($q=1.0$, 1.2, 1.5, and 1.8) were evaluated with five independently generated data sets and matched training seeds, giving 130 held-out cases per regime and 500 posterior draws per case. At $q=1.5$, q-Gaussian weights with a Gaussian likelihood reduced log-conductivity CRPS by 2.85% (95% paired CI 0.82–4.88) and calibration error by 3.72 percentage points. Conductivity RMSE was unchanged, while 95% interval width increased by 9.86%. At $q=1.8$, assigning q-Gaussian tails to both components produced no stable accuracy gain and increased interval width by 138.46%. The Gaussian-generated porosity and dispersion controls showed negligible accuracy changes. Moderate weight-side tails can thus improve distributional prediction, whereas an overly heavy output likelihood trades sharpness for coverage. Separating these two roles provides a clearer basis for uncertainty modeling in groundwater inversion.

Read the full abstract

Medical Physics

Omni-iEEG Planner: an open-source Python platform integrating multimodal neuroimaging and intracranial EEG data for epilepsy surgery research

Yuta Tanoue, Masaki Izumi, Yipeng Zhang et al. (+8 more)

Multimodal intracranial EEG (iEEG) research for epilepsy surgery combines anatomical and functional imaging, implanted electrode locations, interictal biomarkers, and functional mapping–outputs produced by different applications and reconciled by hand before they can be viewed on the same anatomy. Omni-iEEG Planner is an open-source Python platform that brings these inputs into one patient coordinate frame for research data analysis. The workflow starts from the preoperative T1-weighted MRI, from which FreeSurfer provides the cortical surface and anatomical labels; CT, PET and diffusion MRI are then co-registered to that reference, electrode contacts are detected automatically on the postimplantation CT and reviewed by the analyst, and tractography is reconstructed within the application. Channel-indexed results computed outside the platform, such as high-frequency oscillation counts from pyHFO or time-resolved gamma responses, are projected onto the contacts and the cortical surface with their source units preserved, and interictal and functional evidence remain separate layers. We describe the design and demonstrate representative workflows on de-identified patient data. The platform is not proposed as a method for every laboratory; it is a worked example of how one research group combined multimodal integration with an AI-assisted iEEG analysis tool, pyHFO, in a single open environment that others can adapt to their own pipelines. The design will accommodate further evidence types–model-derived contact scores, virtual implantation and hypothetical resection studies, group-level analyses–as they are implemented. The software is released under the MIT license for research and educational use, not clinical decision-making.

Read the full abstract

All authors Yuta Tanoue, Masaki Izumi, Yipeng Zhang, Chenda Duan, Tonmoy Monsoor, Christopher A. Chan, Joe X. Qiao, Noriko Salamon, Aria Fallah, Vwani Roychowdhury, Hiroki Nariai

Computer Vision

Beyond the Last Layer: Global Inter-Layer Aggregation for AI-Generated Image Detection

Wenbin Shen and Zhichao Lian

The rapid development of generative models has made AI-generated images increasingly close to real images. It has also brought several challenges to detectors, including rapid generator updates, large changes in data distribution, and limited cross-dataset generalization. Many existing methods based on pretrained vision encoders primarily use the global representation from the last layer, potentially leaving useful information from intermediate layers underexplored. To address this issue, we propose Global Inter-Layer Aggregation (GILA). GILA extracts global representations from different depths of a frozen encoder, models interactions among layer representations through inter-layer self-attention, and adaptively aggregates the resulting contextualized representations. The method is not tied to a single pretraining objective and does not require fine-tuning the pretrained encoder. We evaluated GILA on three vision backbones with different pretraining objectives, ten seen generators, and three external datasets. The experimental results show that GILA makes better use of multi-layer information in pretrained encoders and improves detection performance and cross-dataset generalization.

Read the full abstract

Image and Video Processing

Coarse-to-Fine Diffusion with Residual Prior for Low-Dose CT Denoising

Ping Wu, Pengcheng Zheng, Jiahao Zhang et al. (+3 more)

Low-dose CT (LDCT) reduces radiation exposure but often introduces severe noise and streak artifacts that obscure subtle anatomical structures. Existing diffusion-based denoising methods commonly rely on implicit assumptions about the noise distribution, which may not hold in practical LDCT scenarios and can degrade denoising performance. To address this limitation, we propose a Residual-Guided Diffusion framework that effectively incorporates an explicit difference prior to inform the diffusion process of noise distribution. The difference prior is computed as the residual between the LDCT image and a coarse clean estimate, constraining diffusion sampling to a plausible residual space. We further adopt a coarse-to-fine strategy in which a pre-trained RED-CNN provides the coarse estimate to anchor global anatomy, and the diffusion model refines the remaining details under the prior guidance. For efficiency, we employ accelerated sampling with 20 steps at inference. Experiments on three representative datasets demonstrate that our approach outperforms existing methods in both quantitative and qualitative performance measures.

Read the full abstract

All authors Ping Wu, Pengcheng Zheng, Jiahao Zhang, Bohao Chen, Yazhou Ren, Xiaorong Pu

Machine Learning

Output-Aware Block Influence: Jacobian-Lens Weighting for Depth Pruning

Kyle Lemoi, Jonathan Yong, Shubhankar Tripathy et al. (+3 more)

Depth pruning removes entire transformer blocks and therefore depends on a reliable criterion for identifying which blocks can be removed. Block Influence (BI) ranks blocks by how minimally they rotate the hidden state. However, local redundancy does not necessarily imply downstream irrelevance. We address this limitation using the Jacobian Lens (J-Lens) neural artifact, a map transporting perturbations at a layer to the final residual stream. At a common sparsity level, pruning scores applying J-Lens transportation to $BI$, what we call $J-BI$, were found to improve post-pruning accuracy by up to 19 points over $BI$ while reducing damage to perplexity for Qwen3-8B, Llama-3.1-8B-IT, and Gemma-3-12B-IT. We also perform a single-block ablation study and use J-Lens probing to explain why the proposed scores improved performance.

Read the full abstract

All authors Kyle Lemoi, Jonathan Yong, Shubhankar Tripathy, Priestley Fernandes, Anirban Majumder, Zarreen Reza

General Mathematics

Bornologies and Kuratowski-Painlevé Convergence to the Empty Set Revisited

Basile Beyer de Ryke and Gerald Beer

Let $(X,\tau)$ be a noncompact metrizable space and let $\langle T_n\rangle$ be a sequence of nonempty closed sets converging to the empty set in the sense of Kuratowski–Painlevé. The subsets of $X$ whose closures meet at most finitely many terms form a bornology $\mathcal B_\tau(\langle T_n\rangle)$. We give an internal criterion for inclusion between two such bornologies and recover each of them from a canonical free filter of closed sets. In fact, a nontrivial bornology is sequence-induced exactly when it has the enlargement property and its closed polar has countable character. Our principal result says that the bornology of relatively compact sets is sequence-induced precisely when $X$ is locally compact and separable; in the locally compact case the character of its closed polar is the density of $X$. In the penultimate section, we make a clever use of the sequence $T_n=\{n\}$ in the real line to give a negative answer to a conjecture posed in 2001 about a coarsest topology making the associated bornology a nontrivial metric bornology. Finally, we end the paper with a more general result: a coarsest such supertopology exists if and only if the original topology already makes the induced bornology a nontrivial metric bornology, in which case the coarsest supertopology is the original topology itself.

Read the full abstract

Multimedia

Beyond Structural Unification: A Perspective and Roadmap on Unified Multimodal Models

Yinyi Luo, Hayes Bai, Yiqiao Jin et al. (+5 more)

Unified Multimodal Models (UMMs) emerge as a promising way to combine understanding and generation within a single architecture. We argue that current approaches mainly achieve structural unification, which can create a Unification Paradox: shared interfaces, token spaces, and parameters do not guarantee an overall capability gain and may degrade individual capabilities. In this position paper, we identify five critical layers in which unification must be rethought: (1) Asymmetric Input Unification, where models must estimate the utility and reliability of each modality rather than treating all modalities equally; (2) Factorized Representation Unification, where semantic abstraction and fine-grained reconstruction should use separate but coordinated capacity; (3) Stateful Reasoning Unification, where models should maintain and revise a shared multimodal belief state across understanding and generation; (4) Synergistic Optimization Unification, where conflicting understanding and generation objectives require coordinated optimization; and (5) Multi-Centric Systemic Evaluation, where benchmarks must evaluate systemic competence, capability preservation, and cross-task robustness rather than isolated outcomes. We argue that progress requires moving beyond structural unification toward functional unification across the full system.

Read the full abstract

All authors Yinyi Luo, Hayes Bai, Yiqiao Jin, Wenyue Hua, Srijan Kumar, Marios Savvides, Qingsong Wen, Jindong Wang

Computer Vision

Task-Utility IQA: Rethinking Image Quality Assessment in Embodied Tasks

Jirong Zha, Yanwei Liao, Yemin Wang et al. (+5 more)

Image quality assessment (IQA) underpins embodied imaging pipelines by judging whether visual quality satisfies downstream tasks, yet most existing methods learn task-agnostic scores aligned with generic human ratings on static benchmarks. This objective mismatches embodied and interactive settings, where image adequacy depends on task goals, context, and action requirements that shape an agent's decisions. We argue that IQA should shift from score regression to goal-conditioned judgment defined by embodied task utility. To this end, we propose an MLLM-based embodied IQA agent framework that reasons about task requirements, grounds quality evidence, and produces decision-facing assessments. A compact VisDrone evidence study shows that conventional perceptual metrics and generic MLLM-based IQA scores are weak proxies for downstream VLM task success, while a task-conditioned utility score better predicts whether visual input supports the target task. These findings support a task-utility-oriented reframing of IQA for embodied, explainable, and decision-aware visual perception.

Read the full abstract

All authors Jirong Zha, Yanwei Liao, Yemin Wang, Xiangmin Yi, Siqi Peng, Yingfeng Chen, Chen Gao, Xinlei Chen

Computer Vision

ObjectTwin: Measuring the Asset Correspondence Gap for Capture-Anchored Digital Twins

Zhenghe Guo

Object-wise reconstruction increasingly inserts generated meshes into captured scenes as editable proxies for observed objects. This use creates a requirement absent from isolated-asset scores and pose estimates: at the captured metric state, the proxy should reproduce calibrated observations of the same instance. We present ObjectTwin, a scan-grounded measurement protocol that freezes the instance, cameras, scan-projected support and recorded depth, metric anchor, and evaluation denominator, then replaces only the object representation. The resulting asset correspondence gap measures lost observation agreement relative to scan-derived visible geometry without treating the scan as complete physical ground truth. Across two indoor RGB-D capture domains and two generation backends, visible, anchored generated assets lose 0.302-0.497 MaskIoU under this replacement. Every tested scene-level estimate is positive, while same-input generator preference reverses across capture domains. Coverage and overlap-depth analyses locate disagreement in both visible support and surface placement; the evaluated conservative pose refinement does not systematically recover it. These results establish capture correspondence as a separately measurable interface between generated-asset quality and stronger physical or task-level evaluation.

Read the full abstract

Statistics Applications

TGAge: A Phenotypic Aging Clock Based on Tabular Foundation Models and the Gompertz Law for Assisted Living Facility Residents

Chengzhuo Li, Rongjia Zhou, Shanshan Wang et al. (+5 more)

Biological age captures overall health status beyond chronological age, yet existing biological aging clocks are difficult to apply to assisted living facility (ALF) residents due to scarce epigenetic data. We propose TGAge, a second-generation phenotypic aging clock that estimates risk-equivalent biological age from routinely collected clinical measurements. TGAge employs a tabular foundation model to predict all-cause mortality risks over discrete time-to-death intervals and converts the resulting mortality-risk ratio into biological age acceleration via the Gompertz law. We develop and evaluate TGAge on a large cohort of ALF residents curated from a proprietary, multi-state long-term care database. TGAge outperforms clinical indices and established biological aging clocks in survival prediction. Moreover, TGAge-derived age acceleration is significantly associated with multiple aging-related comorbidities. These findings support TGAge as a practical, interpretable, and highly predictive biological aging clock for vulnerable ALF populations using only routinely collected clinical measurements.

Read the full abstract

All authors Chengzhuo Li, Rongjia Zhou, Shanshan Wang, Runze Yan, Carl Yang, Xiaobo Sun, Huiwen Xu, Jiaying Lu

Quantitative Methods

TIAgent: From Natural Language to Grounded Agentic Workflows for Computational Pathology

Aakash Madhav Rao, Mark Eastwood, Danielle Belgrave et al. (+1 more)

Computational pathology (CPath) analyses increasingly depend on multi-step computational workflows that require substantial programming expertise, limiting their accessibility to pathologists and other domain experts without extensive coding experience. Although large language models (LLMs) provide a promising natural-language interface for constructing such analyses, hallucinated operations and incorrect workflow specifications limit their reliability in scientific settings. We introduce Tissue Image Analysis Agent (TIAgent), an automated CPath workflow generation framework that, given a natural-language prompt specifying a complex analytical task together with an input dataset, automatically constructs a verifiable computational graph employing existing CPath operations. The resulting graph can be inspected and executed, enabling natural-language construction of complex CPath workflows for applications including automated biomarker discovery and tissue analysis. We demonstrate end-to-end workflow generation and execution and systematically benchmark 14 backbone LLMs across 90 unit tasks spanning 18 CPath operation types. The ability of each LLM to generate a complete and correct unit workflow is quantified using Unit Workflow Success (UWS). Performance varied substantially across backbone models, with Kimi-K2.7-Code achieving the highest UWS of 0.967 (95% CI: 0.907-0.989). Analysis of node and parameter correctness further revealed that selecting the appropriate computational operation and correctly configuring it represent distinct challenges for LLM-based workflow generation. TIAgent provides a foundation for making complex CPath workflow construction more accessible while enabling systematic evaluation of LLM-driven scientific workflow generation.

Read the full abstract

All authors Aakash Madhav Rao, Mark Eastwood, Danielle Belgrave, Fayyaz Minhas

Computation and Language

FusionEval: Benchmarking Model Fusion for Large Language Models

Shuo Cai, Yanggan Gu, Yuhang Liu et al. (+8 more)

Model fusion aims to integrate the capabilities of multi-expert models into a single model. Existing evaluations offer limited coverage of reasoning with long chains of thought and tool use, while strong average performance can mask substantial losses in individual expert capabilities. We introduce FusionEval, a benchmark that provides controlled expert pools, integrates fusion methods with and without additional training, and offers a unified evaluation suite. The pools contain domain experts independently trained from Qwen3 and Llama base models across math, code, science, instruction following, and tool use. The suite evaluates capability retention and computational cost on eight public benchmarks under comparable data access rules, reporting normalized average and worst performance, runtime, and peak GPU memory usage. Extensive experiments show that training-free methods preserve reasoning performance but substantially degrade tool use at both model scales. In contrast, MOPD preserves tool use and achieves a more balanced capability profile, but requires additional data preparation, inference, and training. Jointly retaining reasoning, instruction following, and tool use at low computational cost remains a challenge. We hope FusionEval supports future research on efficient model fusion that jointly preserves expert capabilities. Our code is available at https://github.com/Baicaihaochi/FusionEval_Benchmark.

Read the full abstract

All authors Shuo Cai, Yanggan Gu, Yuhang Liu, Su Lu, Yuanyi Wang, Yang Yu, Zhen Li, Zhitian Hou, Guanghao Zhu, Ming Li, Hongxia Yang

Computation and Language

TauIndianBankBench: Procedure-Aware, Action-Oriented Evaluation of Tool-Using Conversational Agents

Krishanu Adhikary and Prashant Devadiga

Many Indian banks serve customers through FAQ pages and scripted chatbots. Assistants based on language models could respond faster and carry out requests such as blocking a card, but only if they give correct information and follow the bank's rules. These rules prescribe a procedure, a set of required steps, such as obtaining a closure quote before closing a deposit. Adapting models to this domain requires measuring how well they follow the procedure, before and after training. Benchmarks usually judge only the records left at the end of a conversation. Many required steps, however, leave the records unchanged, so an assistant that skips them can still be judged successful. We present TauIndianBankBench, an action-oriented benchmark of 1,000 Indian retail-banking tasks where an assistant helps a simulated customer. Besides the final records, it checks that the assistant took every required step and made no unrequested change. For every model, 6.5-12.0% of sessions are judged successful when only the final records are checked, although the assistant skipped a required step or made an unrequested change. Scoring both procedure and outcome shows more accurately how well language models perform in Indian retail banking.

Read the full abstract