# Research Papers — 2026-07 · bundle page 169 of 222 > Full-text markdown of papers #8401–8450 of 11075 published in 2026-07, newest first. AI-readable mirrors of the full 11075-paper 2026-07 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-07/page/169 [← Previous bundle](https://aipod.dev/papers/archive/2026-07/page/168.md) [Next bundle →](https://aipod.dev/papers/archive/2026-07/page/170.md) ### Transfer Learning for Linear Discriminant Analysis with a Shared Classification Signal Authors: Yonghan Zhang, Yimeng Fan, Wenya Luo, Jiang Hu Published: 2026-07-08 Categories: stat.ME, stat.ML arXiv: https://arxiv.org/abs/2607.06936v1 This paper studies transfer learning for linear discriminant analysis in high-dimensional two-class classification. We consider one target domain and several source domains, where the mean difference in each domain is decomposed into a deterministic common component and a domain-specific random deviation. The common component represents a shared classification signal across domains, while the random deviation captures domain-specific heterogeneity. Under spiked covariance models, we derive deterministic limits for the target-domain Gaussian-calibrated error of weighted transfer classifiers under both homogeneous and heterogeneous covariance settings. These limits quantify the effects of the shared signal, domain-specific variation, dimension-to-sample-size ratios, and spike structures on transfer performance. They further lead to oracle transfer weights and consistent data-driven plug-in estimators. We also characterize the intercept bias induced by unbalanced target-domain class sample sizes and provide an asymptotically optimal correction. --- ### Mathematical methods of reinforcement learning Authors: Denis Belomestny, Alexander Gasnikov, Egor Gladin, Alexey Naumov, Artemy Rubtsov, Yuri Sapronov, Daniil Tiapkin, Nikita Yudin Published: 2026-07-08 Categories: math.OC, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2607.06935v1 Reinforcement learning (RL) is increasingly grounded in tools from probability, optimization, and operator theory. This survey organizes the mathematical structures that underpin the design and analysis of modern algorithms in RL. We begin from Markov decision processes (MDPs) and the Bellman operators, emphasizing contraction mappings, monotonicity, and fixed-point theory that yield convergence guarantees and rates for value and policy iteration, and temporal-difference schemes. We then develop the optimization perspective: stochastic approximation and martingale methods, convex duality and the role of regularization linking mirror/proximal methods. Function approximation is treated through linear and non-linear settings, covering stabilization, error decomposition, and sample-complexity via concentration inequalities for dependent data and mixing processes. We further cover off-policy evaluation/learning, constrained RL and constrained MDPs (CMDPs). Throughout we unify algorithmic templates under common operator and variational lenses, highlighting both finite-sample bounds and asymptotic results. Our presentation is intended to provide a unified mathematical entry point for researchers in probability, optimization, and statistics interested in reinforcement learning. --- ### Imputation Meets Clustering: Exploiting Latent Subgroup Structure for Missing Data Recovery Authors: Chuyao Zhang, E Li, Taochen Chen, Yiqun Zhang, Yuzhu Ji, Shuping Zhao, Peng Liu, Yiu-ming Cheung Published: 2026-07-08 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2607.06930v1 Missing data is prevalent in practical applications, making effective imputation an essential preprocessing step for downstream analysis. Real-world datasets often exhibit complex latent structures composed of multiple subgroups with distinct distributions. However, existing methods often overlook such population heterogeneity. Without explicit structural guidance, these methods tend to produce generic estimates that blur subgroup boundaries and lack instance-level fidelity. While incorporating subgroup information offers a remedy, it faces a circular dependency: reliable subgroup identification requires complete data, while data completion is the imputation objective itself. To resolve this, we propose CAGI (Cluster-Aware Generative Imputation), a framework that reformulates clustering and imputation as a mutually reinforcing co-optimization process. CAGI employs a ``Partition-Guide-Restore'' strategy where dynamic cluster assignments act as local priors to condition a Generative Adversarial Network. An iterative feedback loop is established to progressively refine both cluster structures and imputed values toward faithful subgroup distributions. To ensure distributional stability, CAGI further employs a multi-level optimization objective combining instance-level reconstruction with distribution-level regularization. Extensive experiments on 14 benchmark datasets with 15 representative baselines demonstrate the superiority of CAGI. The source code is available at: https://github.com/supercocachii/CAGI --- ### MADB: A Large-Scale Music Aesthetics Dataset with Professional and Multi-Dimensional Annotations Authors: Sirui Zhang, Tianle Wang, Xinyi Tong, Peiyang Yu, Jishang Chen, Liangke Zhao, Haoxin Zhang, Duo Xu, Xin Jin, Feng Yu, Songchun Zhu Published: 2026-07-08 Categories: cs.SD, cs.AI arXiv: https://arxiv.org/abs/2607.06929v1 Music aesthetic assessment is a challenging yet underexplored problem, requiring models to capture fine-grained, multi-dimensional human perceptual judgments. Progress in this area has been limited by the lack of large-scale datasets with structured aesthetic annotations. We introduce MADB, a large-scale dataset and benchmark comprising 9,999 tracks annotated by 30 trained annotators. Each track is rated by around 10 annotators across 10 perceptual dimensions and one overall score, with additional textual comments for multimodal analysis. We establish a unified evaluation framework over multiple pretrained models. Results reveal substantial gaps between model predictions and human judgments, exposing key limitations of current approaches. MADB provides a new benchmark for human-aligned music understanding. Project page: https://github.com/knownree/madb --- ### Grounding Spatial Relations in a Compact World Model: Instruction Leakage and a Goal-Free Dynamics Fix Authors: Yufeng Wang, Lu Wei, Haibin Ling Published: 2026-07-08 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.06925v1 Compact world models that condition on a language goal promise to ground relations such as ``put the red block left of the blue block'' using a sparse set of explicit \emph{reference anchors}. We ask when such references actually ground a relation, and identify a trap: a goal-conditioned predictor reaches a striking $0.90$ relation-readout accuracy, yet this is \emph{instruction transcription}, not perception. Withholding the goal collapses it to chance ($0.90\!\to\!0.27$, three seeds) and a counterfactual instruction makes the predicted anchors follow the \emph{false} instruction $94.5\%$ of the time (true scene $2.3\%$; $N{=}256$). Tested across three settings and a within-task ablation, our central claim characterizes the confound: \textbf{instruction leakage occurs when the scored quantity is transcribable from the instruction (when the instruction names the answer) and is essentially independent of how predictive the non-instruction inputs are.} Our tabletop and the external BabyAI benchmark leak, whereas a Language-Table forward-dynamics world model whose instruction names \emph{referents} does not, until the instruction is augmented to name the direction; and degrading the action never increases leakage, the opposite of what predictor-competition predicts. The diagnosis prescribes the fix: keep the goal out of the dynamics (it belongs to the planner's cost) and supervise the \emph{read} path, recovering genuine, instruction-independent grounding ($0.88$, identical with and without the goal). The detection protocol and remedy apply to any goal-conditioned world model whose instruction names the scored quantity. --- ### Intrinsic-Noise Consolidation: A Doob-Barrier-Conditioned Diffusion Turns Analog Device Noise into a Continual-Learning Resource Authors: Gunner Levi Howe Published: 2026-07-08 Categories: cs.LG, cs.NE arXiv: https://arxiv.org/abs/2607.06924v1 On analog neuromorphic hardware, intrinsic device noise is normally an accuracy tax. We ask whether it can instead consolidate memories. We cast per-synapse consolidation as a Doob h-transform: condition each weight's stochastic dynamics on never crossing a memory-critical barrier around its consolidated value. The conditioned diffusion gains an extra drift sigma^2 d/dw log h, a restoring force amplified by the noise variance itself that diverges at the barrier. We are explicit about novelty: the anchored drift -s(w-mu) our rule also contains is not ours (the limit of OUA, MESU, and EWC), and we surrender it. We claim only the conjunction of (a) the Doob barrier-conditioning as a synaptic rule, to our knowledge unclaimed (every h-transform use we found is generative modeling, none synaptic), and (b) a falsifiable prediction: increasing intrinsic noise non-monotonically improves sequential-task retention, an inverted-U that anchored-drift methods cannot produce. We pre-registered this as a go/no-go gate; it passes. On single-head Split-MNIST (8 seeds) the rule lifts retention 10.9 points at an interior optimum (paired Wilcoxon p=0.004), while matched OU/EWC/MESU anchors are monotone. Ablating the conditioning removes the effect; the optimum tracks the barrier; the inverted-U survives a second task stream and the realization where noise enters the forward pass. We then measure the intrinsic noise on real BrainScaleS-2 silicon (additive, trial-to-trial independent, tunable via on-chip averaging) and run the rule on the chip with its noise in the training loop: barrier-conditioning retains a prior task 15.6 points better than the matched control at matched average accuracy, a stability-plasticity shift, not a net-accuracy win (single seed; retention measured, energy modelled). Intrinsic analog noise thus becomes a consolidation dividend a digital accelerator must spend energy to generate. --- ### Bi-PT: Bidirectional Cross-Attention Point Transformers for Four-Chamber Heart Reconstruction from Sparse Cardiac MRI Data Authors: Chenchuhui Hu, Shaoming Pan, Leon Axel, Meng Ye Published: 2026-07-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.06923v2 We propose Bi-PT, a pipeline for reconstructing 3D four-chamber human heart meshes from clinical sparsely sampled cardiac magnetic resonance imaging (CMR) data. This work addresses the error-prone generation of 3D cardiac shape from a sparse point cloud (SPC) extracted from 2D long-axis and short-axis views used in routine clinical CMR protocols. Bi-PT enables accurate inference of the four-chamber heart mesh from the SPC by learning robust point features via bidirectional point cross-attention between an atlas and the SPC, together with per-point semantic labels that improve correspondence estimation. We formulate the deformation field as a Neural Ordinary Differential Equation (NODE) parameterized by a per-point affine transformation and translation to deform the atlas toward the target heart shape. By learning such a NODE, we can guarantee the deformation field to be a locally affine diffeomorphic deformation. We also integrate a semantic label loss into the Chamfer distance to encourage label-consistent correspondences and add a smoothness regularization to stabilize and improve the learning of the deformation field. Extensive experiments demonstrate that Bi-PT achieves accurate and robust performance compared to baselines. --- ### Latency-Constrained DNN Architecture Learning for Edge Systems using Zerorized Batch Normalization Authors: Shuo Huai, Di Liu, Hao Kong, Weichen Liu, Ravi Subramaniam, Christian Makaya, Qian Lin Published: 2026-07-08 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2607.06922v1 Deep learning applications have been widely adopted on edge devices, to mitigate the privacy and latency issues of accessing cloud servers. Deciding the number of neurons during the design of a deep neural network to maximize performance is not intuitive. Particularly, many application scenarios are real-time and have a strict latency constraint, while conventional neural network optimization methods do not directly change the temporal cost of model inference for latency-critical edge systems. In this work, we propose a latency-oriented neural network learning method to optimize models for high accuracy while fulfilling the latency constraint. For efficiency, we also introduce a universal hardware-customized latency predictor to optimize this procedure to learn a model that satisfies the latency constraint by only a one-shot training process. The experiment results reveal that, compared to state-of-the-art methods, our approach can well-fit the 'hard' latency constraint and achieve high accuracy. Under the same training settings as the original model and satisfying a 34 ms latency constraint on the ImageNet-100 dataset, we reduce GoogLeNet's latency from 40.32 ms to 34 ms with a 0.14% accuracy reduction on the NVIDIA Jetson Nano. When coupled with quantization, our method can be further improved to only 0.04% drop for GoogLeNet. On the NVIDIA Jetson TX2, we compress VGG-19 from 119.98 ms to 34 ms and even improve its accuracy by 0.5%, and we scale GoogLeNet up from 20.27 ms to 34 ms and achieve higher accuracy by 0.78%. We also open source this framework at https://github.com/ntuliuteam/ZeroBN --- ### Compass: Prostate Cancer Detection Needs Multi-View Context Authors: Paul F. R. Wilson, Mohamed Harmanani, Zhuoxin Guo, Obed K. Dzikunu, Hannes Cash, Adam Kinnaird, Brian Wodlinger, Purang Abolmaesumi, Parvin Mousavi Published: 2026-07-08 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2607.06919v1 Artificial intelligence (AI) analysis of micro-ultrasound ($μ$US) has shown promise for prostate cancer (PCa) detection. However, most existing AI methods focus on the analysis of single $μ$US images in isolation. By contrast, expert $μ$US readers typically assess a full recorded video study, which provides three-dimensional context, to improve PCa detection compared to single-frame analysis. Inspired by this clinical workflow, we propose Compass, a novel AI methodology which models a $μ$US study as a stream of 2D images. Compass jointly integrates rotational sweep videos of the prostate with $μ$US frames acquired at the moment of biopsy, and performs evidence aggregation across the study using a transformer conditioned on the probe's rotational angle. Finally, a decoder head predicts frame-level and study-level risk scores for the patient. The model is trained and evaluated using a multi-center clinical trial dataset of $μ$US studies, including continuous rotational scans of the prostate and videos captured during biopsy acquisition. We compare the proposed method to baseline AI methods from the literature and to risk scores provided by clinical experts. Our framework shows strong performance, highlighting the value of multi-view context for $μ$US PCa detection, and providing a potentially powerful tool to complement human expertise in $μ$US-based PCa diagnosis. Our code is available at: https://github.com/mharmanani/Compass. --- ### LoCA: Spatially-Aware Low-Rank Convolutional Adaptation of Vision Foundation Models Authors: Sojung An, Junha Lee, Sujeong You, Nam Ik Cho, Donghyun Kim Published: 2026-07-08 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2607.06918v2 Pre-trained Vision Foundation Models (VFMs) provide strong visual representations for diverse downstream tasks. The key challenge of VFM adaptation stems from the prohibitive costs of full fine-tuning and catastrophic forgetting. To address this, Low-Rank Adaptation (LoRA) has emerged as the prevailing paradigm for Parameter-Efficient Fine-Tuning (PEFT). However, LoRA is typically designed for transformer self-attention layers parameterized by 2D matrices. Since convolutional kernels inherently couple spatial and channel information within a 4D tensor, forcing them into a monolithic 2D matrix disrupts the inherent spatial topology. In this paper, we propose Low-Rank Convolutional Adaptation (LoCA), a convolution-aware PEFT framework that addresses spatial-channel entanglement by decoupling channel and spatial adaptation. LoCA introduces a low-rank channel adaptation for dense cross-channel mixing and refines spatial bases extracted from pre-trained kernels via Singular Value Decomposition (SVD). Experimental results show that LoCA preserves pre-trained spatial priors and achieves competitive or state-of-the-art performance across fine-grained classification, domain-generalized semantic segmentation, and generative benchmarks. --- ### LoCA: Spatially-Aware Low-Rank Convolutional Adaptation of Vision Foundation Models Authors: Sojung An, Junha Lee, Sujeong You, Nam Ik Cho, Donghyun Kim Published: 2026-07-08 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2607.06918v1 Pre-trained Vision Foundation Models (VFMs) provide strong visual representations for diverse downstream tasks. The key challenge of VFM adaptation stems from the prohibitive costs of full fine-tuning and catastrophic forgetting. To address this, Low-Rank Adaptation (LoRA) has emerged as the prevailing paradigm for Parameter-Efficient Fine-Tuning (PEFT). However, LoRA is typically designed for transformer self-attention layers parameterized by 2D matrices. Since convolutional kernels inherently couple spatial and channel information within a 4D tensor, forcing them into a monolithic 2D matrix disrupts the inherent spatial topology. In this paper, we propose Low-Rank Convolutional Adaptation (LoCA), a convolution-aware PEFT framework that addresses spatial-channel entanglement by decoupling channel and spatial adaptation. LoCA introduces a low-rank channel adaptation for dense cross-channel mixing and refines spatial bases extracted from pre-trained kernels via Singular Value Decomposition (SVD). Experimental results show that LoCA preserves pre-trained spatial priors and achieves competitive or state-of-the-art performance across fine-grained classification, domain-generalized semantic segmentation, and generative benchmarks. --- ### Smart Scissor: Coupling Spatial Redundancy Reduction and CNN Compression for Embedded Hardware Authors: Hao Kong, Di Liu, Shuo Huai, Xiangzhong Luo, Weichen Liu, Ravi Subramaniam, Christian Makaya, Qian Lin Published: 2026-07-08 Categories: cs.CV, cs.AR, cs.LG arXiv: https://arxiv.org/abs/2607.06915v1 Scaling down the resolution of input images can greatly reduce the computational overhead of convolutional neural networks (CNNs), which is promising for edge AI. However, as an image usually contains much spatial redundancy, e.g., background pixels, directly shrinking the whole image will lose important features of the foreground object and lead to severe accuracy degradation. In this paper, we propose a dynamic image cropping framework to reduce the spatial redundancy by accurately cropping the foreground object from images. To achieve the instance-aware fine cropping, we introduce a lightweight foreground predictor to efficiently localize and crop the foreground of an image. The finely cropped images can be correctly recognized even at a small resolution. Meanwhile, computational redundancy also exists in CNN architectures. To pursue higher execution efficiency on resource-constrained embedded devices, we also propose a compound shrinking strategy to coordinately compress the three dimensions (depth, width, resolution) of CNNs. Eventually, we seamlessly combine the proposed dynamic image cropping and compound shrinking into a unified compression framework, Smart Scissor, which is expected to significantly reduce the computational overhead of CNNs while still maintaining high accuracy. Experiments on ImageNet-1K demonstrate that our method reduces the computational cost of ResNet50 by 41.5% while improving the top-1 accuracy by 0.3%. Moreover, compared to HRank, the state-of-the-art CNN compression framework, our method achieves 4.1% higher top-1 accuracy at the same computational cost. The codes and data are available at https://github.com/ntuliuteam/smart-scissor --- ### Seeing What Matters: Lesion-Aware High-Resolution Patch Discovery and Fusion for Chest X-ray Report Generation Authors: Yingshu Li, Yunyi Liu, Zhenghao Chen, Tong Chen, Zailong Chen, Lingqiao Liu, Lei Wang, Luping Zhou Published: 2026-07-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.06909v1 Despite rapid advances in chest X-ray (CXR) foundation models, most radiology report generation (RRG) systems still rely on heavily downsampled inputs (e.g., 256x256) due to the fixed visual token budgets of pretrained vision encoders, suppressing subtle yet clinically important cues present in native-resolution images. However, enabling high-resolution (high-res) perception remains challenging: naive tiling causes prohibitive token inflation, while global compression suppresses subtle lesions and degrades diagnostic fidelity. Inspired by radiologists' workflow, localizing suspicious regions before detailed high-res assessment. We propose Lesion-Aware High-Resolution Patch Discovery and Fusion for Chest X-ray Reporting (LePaX), the first RRG framework that enables efficient high-res CXR perception (up to 1920x1920) without increasing the vision-token count. LePaX formulates high-res perception as a constrained spatial resolution allocation problem under a fixed token budget and introduces two key components: Learnable Spatial Resolution Allocation (LSRA), which learns a spatial utility map that adaptively allocates limited high-res capacity to diagnostically relevant regions, enabling targeted extraction of high-res patches from native CXRs; and Global-Regional Fusion (GRF), which performs token-preserving region-to-global refinement by projecting high-resolution regional evidence back onto the global feature grid through spatially aligned resolution write-back, avoiding token inflation. Experiments on multiple CXR benchmarks demonstrate that LePaX consistently improves both clinical and linguistic metrics while enabling native-resolution CXR perception with over 10x fewer visual tokens than naive high-res tiling. --- ### The Harness Effect: How Orchestration Design Sets the Token Economics of Enterprise Agentic AI Authors: Muayad Sayed Ali, Aliaksandra Novik, Anji Boddupally, Artem Yavorskyi, Chris Nickerson, Daniel Rica, Emily DuGranrut, Felix Leung, Garrett Prince, Grace Barnett, Heath Robinson, Hosain Al Ahmad et al. Published: 2026-07-08 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.06906v1 Agentic AI development today runs on token maxing: buying capability with tokens -- longer reasoning traces, more turns, wider tool payloads, bigger replayed contexts -- so tokens per task grow faster than task value. Falling per-token prices mask the pattern; total spend rises anyway. We argue the decisive lever against token maxing is the harness: the orchestration layer that assembles context, exposes tools, sequences turns, delegates work, and carries enterprise observability and governance. We isolate it with a controlled swap: 22 locked evaluation tasks, six foundation models (Claude Sonnet 4.6, Gemini 3.1, Gemini Flash 3.5, Qwen 3.6, GLM 5.1, Palmyra X6), changing only the orchestration layer -- a frozen conventional production loop versus the Writer Agent Harness. Holding models constant, the harness cuts blended cost per task 41% ($0.21->$0.12), median wall-clock 44% (48s->27s), and tokens per task 38% (14.2k->8.8k), with task-completion quality at parity (0.78->0.81, directional at this sample size). Efficiency is model-invariant -- every model gets cheaper (33-61%) -- while quality gains are capability-dependent: a model's gain correlates almost perfectly with its baseline strength (r=0.99, n=6), a phenomenon we term harness leverage. Quality per dollar rises 82%; task-completions per million tokens rise from 54.9 to 92.0. On this workload the orchestration layer moved cost per task more than the full spread of the model menu did. We formalize token economics at the orchestration layer (including effective input price under prompt caching), detail the six mechanism families behind the effect -- cache-shape discipline to failure-spend governance -- compare six widely used agent systems on the same axes, and argue the harness is the one component whose efficiency multiplies across every model an organization runs -- present and future. --- ### Dynamic Object Detection and Tracking in Construction: A Fisheye Camera and LiDAR Sensor Fusion Model Authors: Yilong Chen, Huili Huang, Yong K. Cho Published: 2026-07-08 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2607.06896v1 Robust dynamic object detection and tracking are essential for enabling robots to operate safely and effectively alongside humans in complex environments such as construction sites. While LiDAR-based SLAM and occupancy grid methods offer viable solutions for detecting and tracking motion, many state-of-the-art 3D vision approaches rely heavily on pre-trained neural networks and require additional post-processing to identify moving objects. Sensor fusion techniques, combining the precision of LiDAR with the semantic richness of RGB imagery, offer a promising alternative. In this work, we present a novel framework that enhances a quadruped robot equipped with a LiDAR sensor and an upward-facing fisheye camera for real-time dynamic object detection and tracking. After identifying moving objects within a registered point cloud, our method assigns semantic labels by projecting 3D coordinates onto a 2D cylindrical panorama, aligning with real-time image-based detections for observation update of the Kalman filter. The proposed system demonstrates high precision, simplicity, and robustness, particularly in handling objects transitioning between dynamic and static states, thus it is well-suited for deployment in real-world construction environments. --- ### Computing with Stochastic Oracles in AI-Augmented Computation Authors: Jie Wang Published: 2026-07-08 Categories: cs.CC, cs.AI, cs.IT, cs.LG arXiv: https://arxiv.org/abs/2607.06893v1 The Stochastic-Oracle Turing Machine (SOTM) framework models AI-augmented computation as the interaction of a probabilistic Turing machine with an oracle whose responses are drawn from context-dependent distributions. This paper studies what an SOTM can achieve under two oracle-response schemes: in a cached-response oracle, each distinct query receives one response that is reused on later calls to the same query, while in a fresh-response oracle, each call returns an independent response. In both schemes, the SOTM first computes from its input and internal random source to generate its first query, then proceeds adaptively, computing from its query-response transcript (the record of queries issued and responses received) to generate each subsequent query or produce a final output. Cached responses impose two transcript-based ceilings on achievable performance: a correct-identification ceiling governed by the total variation distance between the transcript distributions induced by the hidden states of the oracle, and an output quality ceiling equal to the expected score of the best output the SOTM can compute from the transcript. Fresh responses can raise these ceilings by allowing repeated calls to accumulate independent evidence toward correct or high-quality outputs. In the binary single-informative-query case, the error probability decreases exponentially in the number of calls to the same query at the Chernoff rate. For output quality, query-count bounds characterize threshold stopping when the score function is incorporated as part of the SOTM, and majority-based amplification bounds characterize the binary candidate-output model when it is not. Together, the results identify how response reuse, transcript information, and access to the score function determine what an SOTM can compute and at what token cost. --- ### ReMoDEx: A Local-to-Global Relevance-Based Model Decision Explainability Framework for large-Scale Image Datasets Authors: Abhay Kumar Pathak, Mrityunjay Chaubey, Manjari Gupta Published: 2026-07-08 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2607.06889v1 Deep learning image classifiers achieve strong predictive performance yet remain opaque in how decisions are formed. A model may predict correctly while relying on irrelevant cues, shortcut associations, peripheral structures, or device level artifacts instead of task relevant regions. On large scale datasets this opacity is especially problematic, since inspecting heatmaps one sample at a time cannot scale to thousands of predictions. We propose Relevance Based Model Decision Explainability (ReMoDEx), a framework for systematic, dataset scale assessment of model decision behaviour in image classification. ReMoDEx defines a stepwise pipeline: model inference, target class selection, relevance map generation, heatmap standardisation, similarity based grouping of patterns, cluster level interpretation, and spatial relevance assessment. Local methods GradCAM++, Integrated Gradients, Occlusion Sensitivity, and Layerwise Relevance Propagation are each combined independently with a single global module that summarises an entire set of relevance maps into a few decision strategy clusters, replacing sample by sample inspection with an automatic, scalable summary. To demonstrate ReMoDEx, we applied it to a VGG16 based classifier distinguishing COVID-19, Normal, Lung Opacity, and Viral Pneumonia. The classifier showed stable performance (86.27% test accuracy, 0.9624 test AUC). However, each explainer combined with the global module consistently produced two recurring strategies: central thoracic region decisions and border/corner sensitive decisions, indicating possible shortcut learning that conventional metrics could not reveal. Masked image validation confirmed that model confidence and predicted class changed when central or peripheral regions were occluded. ReMoDEx thus provides a scalable relevance based decision assessment framework and an essential complement to accuracy based evaluation. --- ### Converge to Surprise: Evolutionary Self-supervised Image Clustering Authors: Canlin Zhang, Xiuwen Liu Published: 2026-07-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.06887v2 A variety of self-supervised image clustering approaches are invented in the past years. However, all dominant approaches are exploitative: The direction of parameter updates is determined by known states (observed input samples and existing parameters). We propose an explorative self-supervised learning framework that steps out of this zone. We define a surprise score that measures how unlikely the model's output representation is, assuming that all pixels are i.i.d. random noise. Maximizing the surprise score forces the deep learning model to reject the random noise null hypothesis, or equivalently, to discover non-randomness from data. Also, we propose a fundamental assumption: a surprise score cannot, in general, be fully optimized by exploitative optimization approaches. Thus, we propose the converge-to-surprise scheme to optimize a model: an evolution-strategy (ES) outer loop, which maximizes the surprise score using the mutation-selection mechanism, paired with a periodic gradient-descent inner loop, which uses the surprising clusters already discovered by ES as surrogate targets. On simple image benchmarks, our framework trained from scratch achieves new state-of-the-art results in non-parametric self-supervised image clustering --- the strictest deep-clustering setting, where the number of classes is unknown during training. --- ### Converge to Surprise: Evolutionary Self-supervised Image Clustering Authors: Canlin Zhang, Xiuwen Liu Published: 2026-07-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.06887v1 Most self-supervised image clustering models, actually almost all deep learning approaches, are based on gradient descent: In order to calculate the loss, every optimization step requires a clearly defined target, whether a contrastive split, a masked patch or entity, an EMA-teacher output, a pseudo-label, or a differentiable information-theoretic functional. We propose a self-supervised framework that drops this requirement for image clustering. Without any prior knowledge, we have to assume that each pixel is i.i.d. according to the Principle of Maximum Entropy. Taking this as our null hypothesis H0, we define a "surprise score" that measures how unlikely the model's output representation would be under H0. Maximizing the surprise score forces the deep learning model to reject H0 - equivalently, to discover non-random feature from data. Also, here is our fundamental assumption: a surprise score cannot, in general, be reduced to a per-step loss. Hence, we propose the "converge-to-surprise" scheme to optimize our model: an evolution-strategy (ES) outer loop, which directly maximizes the surprise score without needing its gradient, paired with a periodic gradient-descent inner loop, which uses the surprising clusters already discovered by ES as surrogate targets. On standard image benchmarks, our framework achieves new state-of-the-art results in non-parametric self-supervised image clustering - the strictest deep-clustering setting, in which the number of ground-truth classes is not given to the model. --- ### Finding a stationary point of a stochastic convex problem Authors: Felipe Areces, John Duchi, Malo Sommers Published: 2026-07-08 Categories: stat.ML, cs.LG, math.OC arXiv: https://arxiv.org/abs/2607.06883v3 We consider the problem of finding stationary points for stochastic convex optimization problems. Rather than surrogates to stationarity, such as a proximity-to-stationarity guarantee or small gradient of the Moreau envelope, we ask for a stronger notion: that the subdifferential of the objective actually contains a small element. This criterion is non-trivial, because subdifferentials of convex functions fail to converge uniformly, even in arbitrarily small neighborhoods of the optimum. Our convergence guarantees rely on dimension theory to decompose the graph of the subdifferential of a convex function, showing how stochastic sampling preserves "pieces" of these graphs, and allowing effective application of proximal-point-like methods. --- ### GemNav: Discrete-Token Visual Robot Navigation using a Multimodal Large Language Model Authors: Peter Bohm, Saimunur Rahman, Abdelwahed Khamis, Sagun Man Singh Shrestha, Chris McCool, Peyman Moghadam Published: 2026-07-08 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2607.06882v2 Visual navigation policies built on large pretrained models have so far followed a common recipe: a dedicated visual encoder, a bespoke action head, and training on thousands of hours of cross-embodiment datasets. We ask whether this recipe is necessary. In this paper, we introduce GemNav, a visual robot navigation policy that adapts a frozen Multimodal Large Language Model (MLLM) for short-to-medium horizon waypoint navigation using Low-Rank Adaptation (LoRA) on the language tower alone, with no auxiliary visual encoder and no continuous regression head. Waypoints and categorical navigation signals share a single discrete token vocabulary generated by the language-model head, and a soft-decoded auxiliary loss recovers the metric structure that pure cross-entropy training discards. On a single 8.7-hour open corpus, roughly three orders of magnitude smaller than competing training sets, the policy transfers zero-shot to four physically distinct unseen environments and stops within 0.25-0.42m of the goal across 20 real-world trials covering an open carpark, an obstacle carpark, a long outdoor chemical yard, and an indoor warehouse. Conditioning on short image histories improves offline metrics but yields no robot benefit, pointing to a ceiling on what temporal context adds once pretrained vision features are in place. These results indicate that discrete-token adaptation of frozen MLLMs can provide a data-efficient, deployable alternative for foundation model robot navigation. --- ### Best-Arm Identification with Generative Proxy Authors: Tianyi Ma, Hanzhang Qin, Ruihao Zhu, Jierui Zuo Published: 2026-07-08 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2607.06879v1 Best-arm identification is a canonical model for data-driven decision-making, but in many applications each reward observation is costly. Motivated by the growing availability of cheap predictions from machine learning and large language models, we study fixed-confidence best-arm identification in which each costly reward pull is paired with a cheap but correlated proxy score. The marginal mean of the proxy can be estimated offline and is treated as known, whereas its correlation $ρ$ with the reward, which governs how much the proxy helps, is unknown and must be learned online in pair with real rewards. We show that a control-variate adjustment turns this model into a heteroscedastic identification problem whose oracle sample complexity improves by residual variance $1-ρ^2$. The central difficulty is that the correlation must be learned from the same costly samples that identification consumes online, and that a plug-in estimate of the residual variance is anti-conservative and can compromise correctness. We propose PROBE (PRoxy OLS for Best-arm Exploration), a phase-elimination algorithm that directly maintains an upper certificate on the residual variance with an ordinary least squares fit, whose exact chi-square law keeps the certificate valid regardless of the unknown correlation. We prove that PROBE is $δ$-PAC and attains the known-correlation oracle sample complexity up to a constant multiplicative factor and a constant additive calibration cost. The guarantee extends to the $(ε,δ)$-PAC setting under minimal changes to the algorithm. Numerical experiments on synthetic instances and on an auto-loan pricing replay with large language model and tabular proxies confirm that the sample savings of PROBE scale with the strength of the reward-proxy correlation, exactly as the theory predicts. --- ### Video2Reaction: Mapping Video to Audience Reaction Distribution in the Wild Authors: Trang Nguyen, Sidong Zhang, Shiv Shankar, Gauri Jagatap, Deepak Chandran, Andrea Fanelli, Madalina Fiterau Published: 2026-07-08 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2607.06875v1 Understanding and forecasting audience reactions to video content are crucial for improving content creation, recommendation systems, and media analysis. To enable audience reaction prediction and other content engagement applications, we introduce $\textbf{Video2Reaction}$, a multimodal dataset that maps short movie segments to a distribution of $\textit{induced emotions}$ of viewers in the wild, as expressed through social media. $\textbf{Video2Reaction}$ spans more than 10,000 videos and serves as a reliable benchmark as well as a training resource for audience reaction prediction. To enable cost-effective continuous annotations as reactions may change over time, we develop a two-stage multi-agent pipeline using only open-source LLMs, achieving 86% correctness under blind human verification despite the inherently noisy and subjective nature of the task. We establish the first benchmark for video-to-reaction-distribution prediction in the wild and show that pretrained foundation video models fail in zero-shot settings, while finetuning transforms them into state-of-the-art predictors capable of modeling both full reaction distributions and dominant responses from video alone. However, the task remains challenging: even the strongest methods achieve only 77% Top-3 F1 in dominant reaction prediction (LLaVA-Next), highlighting a substantial gap in modeling collective audience reaction. \modification{Dataset and code are available at our project page: https://information-fusion-lab-umass.github.io/video2reaction-bench.github.io --- ### Ensemble Deep Learning Approaches for AI-Altered Video Detection Authors: Laiba Khan, Hung-Mao Wu, Wei Lin, Frank Bi, Yousef Abdelhadi, Joshua Jung Published: 2026-07-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.06872v2 The increasing accessibility of artificial intelligence has led to a rapid rise in AI-generated videos, making it more difficult to distinguish between real and manipulated content. Many existing detection methods rely on a single model and often struggle to generalize across different types of deepfakes. In this work, we developed a multimodal deepfake detection system that combines both audio and visual analysis using an ensemble of models. The system includes AASIST for audio-based detection, and EfficientNet, XceptionNet, and MesoNet for analyzing visual features in video frames. The pipeline takes a video as input, separates the audio, and extracts face frames using MTCNN. Each model produces a score indicating the likelihood of the input being fake. These scores are then combined using ensemble strategies, including mean averaging and stacking. Mean fusion provides a simple and stable baseline, while stacking uses a trained meta-model to learn how to combine predictions more effectively. Results show that while individual models perform well on the datasets they were trained on, their performance drops when tested on more diverse datasets. The ensemble approach helps improve overall robustness by combining predictions from multiple models, leading to more consistent performance across different types of deepfakes. This suggests that using both audio and visual information together is a more reliable approach for deepfake detection. Our results highlight generalization to unseen manipulations as the central open challenge, with average accuracy around 70%. --- ### Geometric Collapse: When Vision Models Fail to Verify Physical Causality Authors: Wentao Zhang, Jinhu Qi, Weiqiang Jin, Yifei Zhang, Chan-Tong Lam, Irwin King Published: 2026-07-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.06871v1 Recent progress in large-scale self-supervised learning has improved dense geometric prediction, but it remains unclear whether such scaling yields inference-time physical plausibility checks. We propose Scrambled Edges, a controlled counterfactual that injects salient edge-like cues while violating surface continuity, illumination coherence, and occlusion ordering. With energy-matched and structure-matched controls, we isolate the effect of unsupported edge evidence from high-frequency energy and edge sparsity. Across CNN/ViT/SSL depth predictors on NYU Depth v2 and KITTI, Scrambled Edges induce up to 3.2x larger deviation from clean predictions than energy-matched noise; additional diffusion and flow-matching depth estimators show attenuated but still significant collapse. The resulting Geometric Collapse propagates globally: even with oracle knowledge of the corrupted region, output-level repair recovers only 47%, with substantial error outside the mask. These findings provide controlled behavioral evidence that current dense predictors lack reliable mechanisms to quarantine physically unsupported edge cues, motivating explicit plausibility scoring and selective cue integration. --- ### Optimal Model Activation Policies for Inference Networks of Large Language Models Authors: Foivos Charalampakos, Md Ibrahim Ibne Alam, Iordanis Koutsopoulos, Koushik Kar Published: 2026-07-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.15992v1 Recent advances in large language models (LLMs) have rendered them necessary for Natural Language Processing (NLP) tasks, and their high inference cost motivates the study of cost-performance trade-offs. In practice, several expert LLMs are used in synergy for inference, either in an ensemble mode or in series, yet without a principled approach on how to best use the available models. An adaptive approach can route simple queries to cheaper LLMs and complex ones to more capable, costly models. However, a clear understanding on how to best leverage available expert models is missing. We introduce inference networks, a graph-based framework, where nodes denote different LLMs, and links denote conditional model activations. The inference network design problem is to determine the best topology, namely the best way to use the models that best addresses the cost-performance trade-off. We start from the basic topology of a series of LLM experts, each of which has a different cost and a different level of expertise, which is captured via model confidence. We formulate the problem of optimal activation of these models so as to minimize the expected inference cost subject to a target performance constraint. For this special class of inference networks, we prove that the optimal activation policy has a threshold structure: query the lowest-cost LLM first, and invoke the more expensive LLM only if the confidence falls below a defined threshold. For discriminative tasks, the optimal policy consists of a set of thresholds, one threshold for each class, while for generative tasks, it consists of a single threshold. We provide a structured method to compute the thresholds, and practical confidence estimation mechanisms for both task types. Experiments with open-source LLMs show substantial cost reductions while meeting the specified performance budget. --- ### The Functionalizer: Lossless Functional Decomposition for Subword Tokenization Authors: Connor Makowski, Willem Guter Published: 2026-07-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.15991v2 Standard subword tokenizers either treat every orthographic variation of a word (such as hello, Hello, HELLO, and Héllo) as unrelated vocabulary entries, which fragments the embedding space, or discard this variation through lossy normalization. We present the Functionalizer, a lossless pre-tokenizer framework that factors orthographic and structural variations into a compositional opcode/operand prefix stream before tokenization: a canonical base token (operand) prefixed by parametric transformation operators (opcodes) encoded in the Unicode Private Use Area. We introduce operators covering casing (CAPITALIZE), diacritics (13 dedicated opcodes), and character repetition (REPEAT, MULTIREPEAT), which are fully reversible. Across natural language and code corpora, the Functionalizer enables complete corpus coverage with significantly smaller vocabularies under unconstrained exhaustion conditions, reducing actual vocabulary slot requirements by up to 19.7%. Downstream evaluations on 98M-parameter GPT-2 models show that the Functionalizer improves Python code syntax validity (9.12% vs. 7.70%) while reducing duplicate n-gram repetition in natural language prose. These findings demonstrate that functional decomposition can be an effective mechanism for vocabulary-efficient, structurally aware language modeling, and motivate further validation at production scale. --- ### PhysMent: An Interactive Approach For LLM Reasoning In Physics Problems Authors: Joseph Chan, Utkarsh Jha, Xiyin Yang, Abhinav Jarajapu, Anik Sahai, Eddie Hu, Robin Jeshua Deepak, Stefano Saravalle, Aditya Shah Published: 2026-07-07 Categories: cs.CL, cs.CV arXiv: https://arxiv.org/abs/2609.13152v1 Large language models (LLMs) perform strongly on static science benchmarks, yet their ability to reason about the physical world through active experimentation remains poorly understood. We introduce PhysMent, a benchmark that evaluates LLM physical reasoning via iterative, toolmediated interaction with a MuJoCo physics simulator. Unlike static benchmarks that supply all quantities upfront, PhysMent requires models to discover information by applying forces, querying object states, advancing time, and modifying scene geometry before answering. The benchmark comprises 105 scenes of classical mechanics, organized across four difficulty regimes (Easy/Hard and Single/Multi), three scene modalities (standard, object creation, hidden objects), and a scene-manipulation category, evaluated with a six-dimensional scoring framework. Results show that current models perform reasonably well on qualitative single-concept tasks (up to 80% accuracy) but degrade substantially on quantitative tasks that demand precise, multi-step experimental procedures: most models fall below 30% on the hardest single-concept category, where the bottleneck is procedural (adaptive multi-step tool use) rather than conceptual load. Across the seven models, accuracy ranges from 25% to 67%, with failures due to premature answer submission, inefficient exploration, and inconsistent grounding in simulator feedback rather than conceptual gaps. --- ### Token Merging for Multilingual Speech Recognition: A Systematic Study Across Model Scale and Fine-Tuning Authors: Dylan Luke Holyoak Published: 2026-07-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.13151v2 Leading multilingual speech recognition models like Whisper transcribe diverse, low-resource languages without language-specific training but are computationally expensive to deploy. Token merging mitigates this inefficiency by dynamically combining redundant features, shortening the sequence length during inference without requiring retraining. In this paper, we systematically evaluate token merging on the Whisper model family across sixteen diverse languages and three different model sizes. We also test how token merging interacts with fine-tuning (DoRA) on low-resource languages. Our findings show that merging tokens increases computational efficiency with almost no loss in transcription accuracy across most low-resource languages and model sizes, and it works even after the model has been fine-tuned. Our results demonstrate that token merging is a highly practical method for making multilingual speech recognition faster and cheaper to deploy. --- ### Adaptive AI: Energy Efficient Multi-exit TinyML on Intelligent Vision Systems at the Edge Authors: Luca Crupi, Lorenzo Lamberti, Alessandro Giusti, Daniele Palossi Published: 2026-07-07 Categories: cs.AR, cs.CV, cs.DC arXiv: https://arxiv.org/abs/2609.11939v1 Traditional TinyML systems for edge devices achieve high accuracy by relying on fixed-depth models that require a constant number of multiply-accumulate (MAC) operations regardless of the input complexity. This approach wastes critical resources in battery-powered Internet-of-Things (IoT) devices and limits the real-time performance of edge cyber-physical systems. Multi-exit execution schemes mitigate these issues and are widely used on high-end devices such as GPUs, but are rarely exploited on edge IoT devices because they require substantial rethinking given their strict memory and computational constraints. We address these aspects by designing and deploying, on an ultra-low-power GWT GAP9 System-on-Chip (SoC), a novel multi-exit computational scheme, demonstrating it on a MobileNetV2 convolutional neural network (CNN) for the ImageNet-100 classification task. Our approach introduces multiple exits at different CNN depths, each with a confidence-based gating mechanism that dynamically and autonomously decides whether to continue or stop inference. Comparing our multi-exit strategy to the standard MobileNetV2 on a GAP9 SoC, we show a 41% reduction in the average computational cost (from 313 MMAC to 185 MMAC), a 29% lower inference time (from 49 to 35 ms), and an energy saving of 24% (from 2.1 to 1.6 mJ per frame). All these improvements come with a ~1% loss in accuracy compared to the full-depth MobileNetV2, which achieves 80.5%. Finally, comparing our adaptable multi-exit scheme with a third-party state-of-the-art adaptive CNN, also deployed on the GAP9, we achieve more than 2x its computational efficiency, increasing it from 8.1 to 17.2 MAC/cycle. --- ### World-Time Compute with Verified Code World Models Authors: James Schwoebel, Ingrida Semenec, Jenia Rousseva, Marcos Ortiz, Collin Overbay, Christopher Klaus, Anderson Edmond, Manish Bhatt, Rome Thorstenson, Jessica Tsai, Martin G. Frasch Published: 2026-07-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.09163v1 LLMs generalize across a domain only after seeing many real, labeled examples, which most domains lack. We study a way to manufacture it cheaply. When a domain's dynamics can be written as code, one template instantiates into many world models: executable, verifiable programs over symbolic state, each an inexhaustible source of exactly-labeled trajectories. Fine-tuning an LLM on trajectories through many such worlds, which we call world-time compute, a training-time analogue of test-time compute, lifts generalization to held-out worlds it never trained on (synthesized world families). Gains are largest where capability is scarcest: +29 points at 0.5B; the largest model's lift is within noise, consistent with saturation. Labels can be trusted because the worlds are verified code: synthesized-then-checked dynamics are exact over 20-step rollouts and answer 10x out-of-distribution probes exactly (100%), whereas per-step LLM and MLP predictors compound error and collapse. Unlike domain randomization, each world is independently authored and verified; a corrupted-label control shows label exactness, not task variety, drives the gains. On real benchmarks (ARC-AGI grids, List Functions, CLRS) the same lever holds as per-world test-time training. On List Functions the harder cross-world form holds: one adapter trained on 128 disjoint worlds reaches 40% on held-out worlds versus 6% for a corrupted-label control (+34 points, CI (29, 39)). The gain is a saturating regularity, not a law: largest for few-step reasoning and small/weak models, fading for long chains, perception-induced tasks, and saturated tasks; cross-task transfer is weak without shared skill. Worlds are authored and served by OpenWorld, a zero-dependency framework (companion paper). Scope: symbolic state; pixel-native domains remain territory of learned models. All code, recipes, and this manuscript regenerate from one repository. --- ### MedTVL: Harnessing Vision and Language for Medical Time Series Classification Authors: Jiexia Ye, Jia Li, Fugee Tsung Published: 2026-07-07 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2608.28605v1 Recent advancements in multimodal learning for medical time series (MedTS) classification highlight the benefits of integrating complementary modalities for clinical decision. However, existing methods typically focus on bi-modal interactions (e.g., time series and text), leaving the tri-modal synergy between time series, vision, and language largely unexplored. Inspired by diagnostic practice synergizing numerical assessment, visual inspection and clinical context, we introduce MedTVL, a text-guided dual-pathway architecture tailored for MedTS classification. Specifically, it synergizes a convolution-based temporal pathway for fine-grained temporal dynamics from raw numerical sequences and a transformer-based visual pathway for holistic morphological structures from time-series-derived images. Such combination of cross-modal and architectural heterogeneity provides a comprehensive diagnostic perspective. To further resolve potential diagnostic ambiguity, both pathways are guided by adaptive medical textual semantics. Finally, a Mixture-of-Experts mechanism dynamically routes each instance to specialized fusion experts, capturing instance-specific reliance on the temporal and visual pathway outputs. In addition, MedTVL supports multimodal contrastive learning to mitigate the clinical label scarcity challenge. Extensive experiments across multiple medical datasets and tasks, spanning supervised, few-shot, and contrastive learning settings, demonstrate the superiority and transferability of MedTVL, highlighting its potential for robust clinical decision support. --- ### SAREF-based Ontology for Distributed AI Workflows across the Edge-Fog-Cloud Continuum Authors: Viorica Rozina Chifu, Tudor Cioara, Vasile Ofrim, Liana Toderean, Ionut Anghel, Laura Daniele, Cornelis Bouter Published: 2026-07-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.26160v1 Nowadays semantic models provide limited support for representing distributed AI workflows and their execution across heterogeneous edge, fog, and cloud environments. Therefore, AI processes and resources are often described using incompatible semantic representations, affecting the interoperability, orchestration, and reuse. To address these challenges, this paper proposes a SAREF-compliant ontology for representing distributed AI workflows across the edge-fog-cloud continuum. We extend the SAREF4SYST ontology with concepts for modeling AI pipelines, executable AI jobs, computational resources, deployment constraints, and communication relationships, providing a unified semantic model of both AI workflows and heterogeneous computing infrastructures. The ontology enables semantic interoperability, automated reasoning, and resource-aware orchestration of distributed AI applications while remaining fully aligned with the ETSI SAREF ecosystem. The ontology is evaluated using proof-of-concept smart grid energy services orchestration scenarios and validated using competency questions showing its ability to support AI workflow deployment, execution reasoning, and workload adaptation across heterogeneous edge, fog, and cloud environments. All competency questions were successfully validated using SPARQL querying and semantic reasoning. Experimental results demonstrate deployment success rates of 90-100% with average orchestration decision times below 80 ms across heterogeneous edge-fog-cloud environments, highlighting its effectiveness on ensuring semantic interoperability for distributed AI orchestration. --- ### Self-Generated Text Recognition: Quality Heuristics, Cross-Task Transfer, and Downstream Bias in LLM Evaluation Authors: Jesse St. Amand, Callum Canavan, Sohaib Imran, Joseph Hewson, Aaron Lutz, Shi Feng, Puria Radmard, Lennie Wells Published: 2026-07-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.26159v2 Self-Generated Text Recognition (SGTR)--the ability of an LLM to identify its own outputs--poses risks to AI safeguards that rely on LLMs as evaluators or monitors: an LLM may recognize outputs from other copies of the same model and make biased judgments or collude outright. Prior work has drawn conflicting conclusions about whether current models possess significant SGTR capabilities. We explain these disagreements by identifying key experimental design choices--which we term operationalizations--that drive divergent results. Evaluating 13-21 models across six presentation operationalizations and four task-domain operationalizations, we find that accuracy varies substantially with evaluation format (pairwise vs individual assessments of text), conversation format (presenting candidate text in user tags vs assistant tags), and the domain of the task used to generate candidate text (e.g., coding vs summarization). We corroborate previous observations that a quality heuristic--models attributing authorship to text they perceive as higher quality--is a dominant confound. We also find that improving a model's SGTR performance via supervised fine-tuning (SFT) on one operationalization can generalize to others, and can increase the model's preference for its own outputs when it acts as a judge in the AlpacaEval framework. Our results suggest that, despite confounds, some models possess practical SGTR capabilities, and that SGTR should be monitored and considered in the design of safety-critical AI applications. --- ### InfoDPP-PAC: Principled Patch Selection for Whole Slide Image Analysis Authors: Prateek Mittal, Ayush Srivastava, Joohi Chauhan Published: 2026-07-07 Categories: q-bio.QM, cs.CV, cs.IT, cs.LG arXiv: https://arxiv.org/abs/2608.23574v1 Each WSI slide contains thousands of candidate tissue patches, while supervision is usually available only at slide level. Existing bag-construction strategies like Uniform extraction and handcrafted heuristics do not control redundancy while attention-based multiple-instance models couple patch importance to a particular downstream classifier, and coreset methods optimise embedding-space coverage without modelling task-relevant patch quality. We introduce InfoDPP-PAC, a principled patch-selection framework that combines teacher-seeded Gaussian process relevance modelling, determinantal log-determinant diversity,submodular greedy optimisation, and a concentration-based adaptive stopping rule. The main theoretical result shows that the log-determinant diversity term used in DPP-style selection is the Gaussian process mutual information between a selected subset and the latent relevance function. We further derive a PAC-style certificate for residual information gain, allowing the number of retained patches to vary by slide rather than being fixed a priori. The empirical study evaluates whether the selected subset is diverse, spatially and morphologically covering, non-redundant, and enriched for the teacher-derived relevance signal. It does not claim end-to-end diagnostic improvement after retraining a downstream MIL model. On 202 HISTAI gastrointestinal whole-slide images, the adaptive rule uses 83.7% fewer patches on average than a fixed full budget while retaining 97.9% of full-budget composite selection quality. At a matched budget, InfoDPP-PAC achieves the highest mean teacher-derived relevance score among fourteen baselines, with diversity and composite scores close to the strongest coreset methods. The results support InfoDPP-PAC as a controlled quality-diversity-cardinality selection framework, rather than as a downstream clinical predictor. --- ### What If AI Carried Her Imagination? Black Girls as Creators in an AI Storytelling Weekend Program Authors: Chun Li, Lauren Brown, Hubert Asare, Shawna Patterson, Dennis Henderson, Ericka Roland, tara Nkrumah, Angela E. B. Stewart Published: 2026-07-07 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2608.16896v1 This paper presents the design and outcomes of a seven-weekend AI storytelling program developed for Black girls aged 10-12. Grounded in Afrofuturism and Black feminist thought, the program adopted AI-enabled counter-storytelling, supported the development of foundational AI literacies, and fostered future-oriented imagination. Activities included brainstorming AI-related topics, developing character and story plots, and delivering collaborative group presentations. Drawing on the analysis of learners' artifacts from the case study, findings show that participants created Afrofuturist narratives rooted in their identities and everyday experiences. At the same time, they developed core AI literacies, including prompt engineering, bias critique, and awareness of data privacy. This program demonstrates that integrating Afrofuturist storytelling with generative AI in informal learning spaces can be a powerful approach for engaging Black girls in computer science education. --- ### Understanding AI Anxiety in the Workplace: A Multimethod Investigation Using Fear Acquisition Theory and the Technology Acceptance Model Authors: Jaroslaw Grobelny, Mateusz Klakus, Kacper Szymański, Teresa Chirkowska-Smolak Published: 2026-07-07 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2608.14609v1 As artificial intelligence (AI) rapidly diffuses and concerns about job displacement intensify, the psychological mechanisms underlying AI job replacement anxiety remain insufficiently understood. Drawing on Integrated Fear Acquisition Theory and the Technology Acceptance Model, the present research investigates whether AI job replacement anxiety can be elicited through vicarious exposure to narratives emphasizing AI-over-human control, and whether perceived usefulness and perceived ease of use of AI moderate this response. Across two studies, we examine AI job replacement anxiety as a response that emerges through vicarious exposure to narratives emphasizing AI agency and human control loss, rather than through direct personal experience of job displacement. Study 1 employed a randomized experiment (N = 316), demonstrating that such exposure increased AI job replacement anxiety. This effect was moderated by perceived usefulness of AI, but not by perceived ease of use, and remained robust after controlling for core self-evaluations. Study 2 (N = 995) replicated the association between perceived AI-over-human control and job replacement anxiety in an observational design and provided convergent evidence for the moderating role of perceived usefulness, supporting the external validity of the findings. Together, the results provide the first causal evidence that perceptual and vicarious processes can trigger AI job replacement anxiety. By shifting attention from structural labor-market conditions to how AI agency is perceived and communicated, this work offers a mechanism-based account of when and why AI-related job fears arise. --- ### From Prediction to Intervention: Personalized Meal-Level Glucose Regulation via an LLM Agent Authors: Mingyu Huang, Weiqing Min, Ying Jin, Yilin Wang, Shuqiang Jiang Published: 2026-07-07 Categories: cs.HC, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.13581v1 Personalized glucose regulation remains a central yet unresolved challenge in precision nutrition, as postprandial glucose response varies substantially across individuals. Existing approaches based on glycemic indices fail to adequately account for such heterogeneity and lack the mechanism to dynamically adjust meals based on personal physiological feedback. In this context, recent advances in LLM-based agents offer a promising direction, as they enable context-aware reasoning and iterative refinement. Inspired by this, we propose a physio-feedback agentic loop, a unified system that integrates individualized absorption modeling with dietary intervention to regulate glucose response. Specifically, we develop a Physiology-Aware Glucose Predictor to model individualized absorption dynamics through a learnable Temporal Physiological Absorption Decay Module. We then construct a Prediction-Driven Two-Stage Meal Optimization Agent that iteratively refines real-world meals using predicted outcomes as explicit feedback. Through extensive experiments on multiple public datasets, we demonstrate that our method not only improves prediction accuracy but also effectively reduces glucose excursions. To the best of our knowledge, this paper marks the first step in integrating physiological learning with an LLM-based agent for personalized glucose regulation. --- ### Jais 2: A Family of Arabic-Centric Open Large Language Models Authors: Mohamed Anwar, Abed Alhakim Freihat, George Ibrahim, Mostafa Awad, Abdelrahman Sadallah, Gurpreet Gosal, Gokulakrishnan Ramakrishnan, Sarath Chandran, Biswajit Mishra, Rituraj Joshi, Ahmed Frikha, Etienne Goffinet et al. Published: 2026-07-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.13580v1 Jais 2 is a family of Arabic-centric large language models developed jointly by MBZUAI, Cerebras, and Inception, designed to advance Arabic-centric language modeling, with strong performance across the Arabic and culturally grounded benchmarks evaluated in this report. The family includes, to our knowledge, the largest open Arabic-centric LLM trained from scratch at 70B parameters, and a competitive 8B-parameter variant among the evaluated open models. A custom Arabic-centric vocabulary enables efficient training and inference. In addition, an optimized architecture and training recipe yield highly compute-efficient training. With a substantially smaller token budget than comparable models, Jais 2 achieves strong Arabic performance on the benchmarks considered in this report and competitive English results. The models obtain leading results among the evaluated open models on OALL2 and AraGen. They also perform strongly on several culturally grounded Arabic benchmarks, including poetry, religion, cuisine, and dream interpretation, as well as in general tasks such as translation and summarization. We release the models in HuggingFace under a commercially permissive license. Jais 2 70B is also released as a chat app on the Web, iOS, and Android; it runs on Cerebras hardware, delivering up to 2,000 tokens per second, and enabling high-throughput Arabic-centric chat serving in our deployment setting. By uniting scale, linguistic diversity, cultural fidelity, openness, and speed, Jais 2 provides an open-weight foundation intended to support further research and development in Arabic-centric LLMs. --- ### An adaptive and evolvable deep reinforcement learning framework for weather prediction Authors: Qiang Wu, Han Li, Jianping Huang Published: 2026-07-07 Categories: physics.ao-ph, cs.LG arXiv: https://arxiv.org/abs/2608.09948v1 No single AI weather model excels at all variables, pressure levels, and lead times. Rather than building yet another architecture, we reframe the forecasting problem as one of coordination. Here we present Feitian Adaptive Ensemble Weather (FTAE-Weather), a lightweight framework that learns, through deep reinforcement learning, when and where to trust each member of an open pool of pretrained forecasters. A tactical Weight-Agent reads the current atmospheric state and assigns variable- and horizon-specific fusion weights, while a strategic Evolve-Agent periodically prunes underperforming models and absorbs newly released ones. Asynchronous prediction caching keeps training cost independent of the slowest constituent model. Adding fewer than 0.01 percent extra parameters, FTAE-Weather reduces RMSE by from 17.2 percent to 78.3 percent over the best individual model in 10 atmospheric variables and outperforms conventional ensemble baselines across lead times from 72 to 360 hours. The framework thus converts a growing, fragmented inventory of specialist models into a single prediction system that strengthens as the field of AI weather forecasting releases new architectures-turning model diversity from a coordination challenge into a compounding scientific advantage. --- ### Geographically Weighted Surrogate Models for Rapid Small-Area Chronic Disease Estimation Authors: Aanya Gupta, Szandra Péter, Sara Von Hoene, Emma Von Hoene, Taylor Anderson Published: 2026-07-07 Categories: q-bio.QM, cs.LG, stat.AP arXiv: https://arxiv.org/abs/2607.28655v1 Small-area estimation (SAE) enables researchers and policymakers to identify spatial disparities in health outcomes, but survey-based SAE products carry an inherent lag. Gold-standard estimates such as CDC PLACES are released roughly two years after the underlying survey data are collected, limiting their use for time-sensitive decision-making. This study evaluates the potential for machine learning (ML) to serve as a surrogate, learning the relationship between frequently updated area-level predictors and existing SAE outputs to generate timely, comparable estimates in years when SAE from surveys are unavailable or delayed. We evaluate several global and geographically weighted ML models for county-level SAE of ten chronic conditions across the US: COPD, asthma, heart disease, arthritis, cancer, depression, diabetes, high blood pressure, high cholesterol, and stroke. Our findings suggest that geographically weighted ML frameworks like geographically weighted random forest and geographically weighted regression offer scalable and open data surrogates for rapidly generating SAE and supporting data driven decision making. --- ### Psych-ECA: A Reproducible Semi-Synthetic Benchmark for Synthetic Control Arms in Longitudinal Psychiatry Authors: Aakash Bhagat, Shashank Choudhary Published: 2026-07-07 Categories: stat.AP, cs.LG arXiv: https://arxiv.org/abs/2607.27224v1 External and synthetic control arms (ECAs) are entering psychiatric drug development, but the field lacks a benchmark that evaluates the properties regulators care about: not only how accurately a method reconstructs untreated trajectories, but whether its uncertainty is calibrated, whether it is robust to the informative observation times common in mental-health records (sicker patients are seen more often), and what false-positive rate it induces in go/no-go trial decisions. Real psychiatric trial data (e.g. STAR-D and registry cohorts) require credentialed access and lack ground-truth counterfactuals, so, following established semi-synthetic benchmarks in causal inference (IHDP, ACIC, and the PK-PD tumor-growth simulator), we release Psych-ECA, a fully reproducible generator of longitudinal symptom trajectories for depression (PHQ-9), anxiety (HAM-A), and psychosis (PANSS) with known counterfactual control arms, informative visits, and validated-scale measurement noise. We benchmark eight estimators spanning carry-forward, pooled real-world-data averages, nearest-neighbour matching, linear mixed models, gradient boosting, and the Scribe trajectory-bridge method. Three findings emerge. First, trajectory and flexible machine learning methods achieve the best counterfactual accuracy (about 2.3 PHQ-9 RMSE), outperforming cross-sectional baselines. Second, only Scribe is both accurate and calibrated, achieving 93-96% empirical coverage of nominal 90% prediction intervals, compared with 87-88% for gradient boosting and 62-75% for uncalibrated SDE models. Third, inverse-intensity correction reduces bias under informative sampling, while Scribe's calibrated intervals are the only trajectory method that maintains nominal false-positive rates as informativeness increases. We release all code, data-generation scripts, and random seeds to enable fully reproducible evaluation. --- ### Do Modules Stay in Their Lane? Role Drift in Compound LLM Systems Authors: Xiaoyang Cao, Siddarth Srinivasan, Michiel A. Bakker Published: 2026-07-07 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2607.21627v1 End-to-end reinforcement learning can improve the accuracy of compound LLM systems, but it does not constrain how modules divide labor internally. We identify Role Drift, a failure mode in which modules preserve or improve end-task performance while deviating from their assigned roles through role-violating shortcuts that remain invisible to system-level evaluation. To make role drift observable and controllable, we propose Role Anchor, a regularizer that modulates how much each module deviates from its assigned role during end-to-end training. The key idea is to preserve how the role prompt shifts the module's next-token predictions relative to a neutral prompt, which serves as a proxy for the role's intended effect during training. Experiments on two compound LLM pipelines reveal role drift that accuracy alone fails to detect: a decomposer meant to split a question into sub-questions for a separate solver instead plants the answer in them, and a reader meant to answer from retrieved passages instead falls back on parametric memory. In fact, on the decomposer pipeline this shortcut drives most of the apparent RL gain: 86% of it vanishes once the decomposer is held to its role, indicating that terminal accuracy alone can badly overstate how much a compound system has genuinely learned. Across both pipelines, Role Anchor mitigates role drift at a tunable accuracy cost that varies by pipeline and anchor strength. Additional gradient analysis suggests that the regularizer reduces alignment with the role-drift direction rather than simply suppressing learning. --- ### Discrete Action Space as a Prerequisite for GRPO Convergence in Small-Model Continuous Control Authors: Dmytro Filatov, Valentyn Fedorov, Vira Filatova Published: 2026-07-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.21626v1 We study whether Group Relative Policy Optimization (GRPO) can fine-tune small language models for simulated quadrotor continuous-control tasks. In our benchmark, vanilla GRPO fine-tuning of Qwen-0.5B for 25 Hz quadrotor velocity control collapses to the trivial zero action: 0 percent success rate, with entropy falling from 0.35 to 0.03 within 60 steps. Two ablations - removing the jerk-penalty term and removing the KL anchor to the pretrained prior - each prevent entropy collapse, yet neither enables learning. When the action interface is replaced by a 5-way categorical choice over PID presets, training converges. The resulting controller traces a smoothness-reliability Pareto frontier along training duration; both endpoints are reported: 98.6 percent success with 0.656 m/s3 jerk at 64 steps, and 100 percent success with 1.103 m/s3 jerk, or 0.796 under a matched velocity cap, at 256 steps. The recipe is evaluated across three pretrained language models. As context, a re-tuned classical baseline, PID with Ki = 0.30 and vmax = 2.5, reaches the same 100 percent success rate at jerk 0.736 m/s3. A high-fidelity simulation using Crazyflie 2.1 dynamics surfaces a hover-region training-distribution gap. --- ### Trajectory-Aware Retrieval Agents for Temporal Decision- Making Authors: Jing Wang, Jie Shen, Xing Niu Published: 2026-07-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.21625v1 We study the problem of decision-making from long-form, temporally structured text using large language model (LLM) agents. Standard retrievalaugmented generation (RAG) pipelines fragment chronological context into isolated snippets, discarding the temporal structure that is often critical for correct downstream decisions. We introduce TLM (Trajectory Language Model), a closed-loop agentic framework that iteratively refines the evidence set using SHAP-guided feedback. The key technical contribution is the latent growth curve model (LGCM) over retrieved chunk embeddings, which provides an interpretable mechanism for detecting trajectory trends, turning points, and information gaps. We show that, under a scorer-calibration assumption (which holds approximately in practice), the iterative refinement procedure is monotonically non-decreasing in the probability assigned to the correct label. Empirically, TLM is evaluated on three temporally grounded decision tasks: medical question answering, earnings call surprise prediction, and overnight stock gap prediction. TLM substantially outperforms both zero-shot LLM baselines and standard retrieval-augmented approaches on the medical task, and yields consistent, economically meaningful gains on the two financial tasks. --- ### FBLayout: Optimizing Memory Layout for Efficient LLM Finetuning on Mobile GPUs Authors: Kahou Tam, Wei Niu, Yu Bao, Xiaomin Ouyang, Chengzhong Xu, Li Li Published: 2026-07-07 Categories: cs.AI, cs.DC, cs.LG arXiv: https://arxiv.org/abs/2607.21624v1 Transformer-based models have enabled unprecedented capabilities across language, vision, and multimodal tasks. On-device fine-tuning of transformer models offers a privacy-preserving path to personalized AI, yet remains inefficient on mobile GPUs due to severe memory constraints and frequent layout transformations in attention mechanism during training. Existing mobile training frameworks either use unified layouts for forward and backward passes -- leading to fragmented memory access and poor GPU utilization during backpropagation -- or rely on explicit layout conversions, which introduce significant transformation overhead. To overcome this, we propose FBLayout, a layout-aware framework that co-designs tensor organization with mobile GPU platforms. FBLayout introduces: (1) a unified R-Tile layout for multi-dimensional reductions across forward/backward passes; (2) tile-based index transformation to eliminate physical data movement; and (3) activation-guided layout selection to propagate efficient layouts globally. Evaluations on seven transformer models across different mobile phones (including ARM Mali and Qualcomm Adreno GPUs) show that FBLayout achieves 2.2-5.7x speedup over MNN, TFLite, and TVM, while significantly improving cache efficiency and reducing memory footprint, enabling practical on-device large model fine-tuning. --- ### Multimodal CoLRAG-TF: Triple-Filtered Retrieval for Complex PDFs Authors: Takato Yasuno Published: 2026-07-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.20517v1 Retrieval-augmented generation (RAG) over heterogeneous PDF collections remains challenging due to multimodal content, domain-specific terminology, and the need for multi-hop reasoning across dispersed evidence. We present Multimodal CoLRAG-TF, a four-axis fusion architecture that integrates dense text embeddings, BM25 keyword matching, knowledge-graph triple filtering, and image-based similarity for robust retrieval over complex documents. Our system constructs a multimodal index of 2,403 blocks extracted from 43 Japanese disaster lesson PDFs, supported by a hybrid OCR pipeline and LLM-based caption generation. To enhance compositional reasoning, we extract 11,414 OpenIE triples and index them with FAISS, enabling sub-second triple lookup and hierarchical propagation of relevance signals. A HippoRAG2-inspired coarse-to-fine retriever (volume $\to$ chapter $\to$ block) narrows the search space before final fusion scoring. Bayesian optimization over fusion weights reveals that the triple axis must dominate ($α_\text{triple} = 0.44$) to counteract lexical bias and sustain multi-hop retrieval quality. Evaluated on a 457-pair benchmark, Multimodal CoLRAG-TF achieves a Retrieval Recall of 0.9909 and a 71.6$\%$ improvement in multi-hop answer similarity over single-hop queries. An image-to-lesson pipeline using a vision LLM further demonstrates the applicability of the approach to visual inputs. These results show that triple-filtered multimodal fusion is essential for structured reasoning over noisy, heterogeneous PDFs and provides a general framework applicable beyond the disaster domain. --- ### Scaling Closed-Loop Feature Channel Configuration with LLMs Authors: Tolgay Atinc Uzun, Radu Timofte, Dmitry Ignatov Published: 2026-07-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2607.20516v1 Promising initial results in closed-loop large-language-model-based channel-configuration search demonstrated that neural-network widths can be optimized directly through executable code generation and accuracy feedback. However, those results were obtained from a relatively sparse set of valid evaluations, leaving open whether the observed optimization behavior transfers to a denser sampling regime and whether additional architectural regularities emerge when more generated networks are evaluated. To test this, the same search setting is scaled to 250 candidate networks per fine-tuning cycle. The analysis covers 2000 generated candidates from 8 complete cycles, yielding 462 verified CIFAR-100 evaluations after task and metadata filtering. Per-cycle mean accuracy exhibits a positive linear trend with slope 9.87e-4 (p=0.043), while the high-performing frontier improves more strongly: the best observed accuracy increases from 0.3144 to 0.3676, and both the top-5 and top-10 cycle-level means exhibit positive trends. The scaled run also reveals improved parameter efficiency. The best model reaches 0.3676 with 11.8M parameters, compared with an early high-performing model at 0.3144 with 166.5M parameters. Beyond accuracy, the larger sample exposes architectural regularities that were difficult to assess from sparse observations. Non-power-of-two channel widths occur in 41.8% of verified candidates, and the strongest models share structured channel-allocation patterns characterized by moderate early widths and expanded middle or later blocks. These findings indicate that the channel-search signal observed in the initial study transfers --- ### Reliability-Aware LLM Alignment from Inconsistent Human Feedback Authors: Jingyi Huang, Ruohan Zong, Yujun Feng, Liran Ma, Lanyu Shang, Yang Zhang Published: 2026-07-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.20515v1 Reinforcement Learning from Human Feedback (RLHF) is critical for aligning Large Language Models (LLMs) with human preferences. However, its efficacy is often compromised by the inherent inconsistency and subjectivity of human annotations. Existing preference optimization frameworks, such as Direct Preference Optimization (DPO), typically treat ambiguous pairs with high annotator disagreement identically to those with unanimous consensus, forcing models to overfit to inconsistent supervision signals and leading to suboptimal alignment. In this work, we propose Reliability-Guided Preference Optimization (RGPO), a robust framework designed to mitigate the impact of inconsistent human feedback. RGPO estimates annotator reliability and infers latent ground truth labels from noisy human feedback to identify robust preferences. Furthermore, we introduce a reliability-aware consistency optimization that dynamically modulates the training objective based on the consensus level of annotations, ensuring the model prioritizes high-consensus supervision signals. Extensive experiments on LLM alignment benchmarks demonstrate that RGPO effectively reduces inconsistency and noise in training data and achieves superior performance compared to widely adopted RLHF baselines. Our code and configurations are available at https://github.com/GenieHuang/RGPO. --- ### ITPEval: Benchmarking Formal Translation Across Interactive Theorem Provers Authors: Jiayi Wu, Robert Joseph George, Anima Anandkumar Published: 2026-07-07 Categories: cs.AI, cs.LO arXiv: https://arxiv.org/abs/2607.19407v1 Formal theorem proving has emerged as a frontier challenge for machine learning, yet the ecosystem is fragmented: proofs remain siloed across incompatible systems, limiting both training data for learning-based provers and the portability of verified results. We present ITPEval, the first benchmark for evaluating automated formal proof translation across four major ITPs (Lean 4, Rocq, Isabelle, and HOL Light), spanning two distinct logical foundations. Our benchmark comprises 1,560 source files and 6,848 theorems organized into a controlled tier of axiomatized files that isolates foundational translation difficulty, and an ecosystem tier drawn from real libraries that exposes API and proof-style mismatches. We release itpeval, a unified multi-ITP verification infrastructure with state-isolated warm backends that preserve per-artifact native checking semantics. We evaluate both statement and proof translation across five frontier and open-weight LLMs on 12 directed translation pairs: statement translation peaks at 29.1% pass@1 and proof translation at 10.5%; controlled theorems reach 29.7% proof pass@1 versus 5.2% for ecosystem-level translations, confirming that library mismatch is the dominant bottleneck. In addition to pass@k evaluation, a deterministic Lean 4 BEq check establishes equivalence for 54.0% of verified source-to-Lean 4 miniF2F statement translations, showing that native type-checking alone can substantially overestimate semantic fidelity; in an autoformalization/auto-informalization round-trip study, Rocq and HOL Light are easier formalization targets than Lean 4 and Isabelle, while multi-ITP context improves pooled Lean 4 success from 4.8% to 10.6%. Our benchmark, verification infrastructure, and evaluation pipelines are publicly released. ---