# Research Papers — 2026-06 · bundle page 34 of 260 > Full-text markdown of papers #1651–1700 of 12993 published in 2026-06, newest first. AI-readable mirrors of the full 12993-paper 2026-06 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-06/page/34 [← Previous bundle](https://aipod.dev/papers/archive/2026-06/page/33.md) [Next bundle →](https://aipod.dev/papers/archive/2026-06/page/35.md) ### Neuromorphic Energy-Aware Learning for Adaptive Deep Brain Stimulation Authors: Binh Nguyen, Colleen Josephson, Mircea Teodorescu, Gert Cauwenberghs, Jason Eshraghian Published: 2026-06-26 Categories: cs.NE, cs.AI, cs.LG, eess.SY arXiv: https://arxiv.org/abs/2606.28600v1 Neuromorphic and edge computing research has focused on reducing the inference cost of neural network controllers, yet in physical closed-loop systems the actuator can rival or exceed an efficient controller in energy. An efficient controller is therefore necessary but not sufficient, because the actuator becomes the cost worth reducing once inference no longer dominates it. Here, we introduce energy-aware learning, an approach that incorporates actuator energy directly into the reinforcement learning reward, and demonstrate it in closed-loop deep brain stimulation (DBS) for Parkinson's disease. A deep spiking Q-network, trained in a biophysical cortico-basal ganglia-thalamic circuit model, learns to suppress pathological alpha-beta oscillations by 45.2% while reducing stimulation charge by 80.0% relative to continuous DBS. Sparsity-constrained knowledge distillation compresses the policy onto the SynSense XyloAudio 3 neuromorphic processor at 0.52 mW inference power, yielding 28.1x lower energy per inference than an equivalent artificial neural network on conventional edge hardware. By co-optimizing stimulation energy and inference efficiency, the framework addresses both major power demands in implantable neuromodulation. --- ### Conformal Prediction with Macro-Coverage Guarantees Authors: Aabesh Bhattacharyya, Tiffany Ding, Rina Foygel Barber Published: 2026-06-26 Categories: stat.ME, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2606.28598v1 Prediction sets should have high coverage to be useful, but some coverage notions are more practically relevant than others. In the classification setting, class-conditional coverage requires that the prediction set (i.e., the set of candidate labels for a new test point) must achieve the target accuracy level within each class, which may be challenging to satisfy when many classes are rare and have few calibration points. At the other extreme, marginal coverage requires only that coverage holds on average over the distribution of all classes, which can lead to low-probability labels being essentially ignored. To find a middle ground, recent work has introduced macro-coverage, defined as the unweighted average of class-conditional coverages. Macro-coverage offers a compromise between marginal coverage and class-conditional coverage that is particularly appropriate for long-tailed settings. In this work, we show that label-weighted conformal prediction can be used to produce prediction sets with a finite-sample macro-coverage guarantee, and more generally a guarantee on a family of generalized macro-coverage objectives that aggregate coverage at the level of arbitrary class groupings and take a weighted average. We further characterize the form of the smallest prediction sets satisfying a given generalized macro-coverage objective and propose a corresponding conformal score function. We validate our theoretical results on two large-scale image classification datasets. --- ### Animation2Code: Evaluating Temporal Visual Reasoning in Video-to-Code Generation Authors: Anya Ji, Abhijith Varma Mudunuri, David M. Chan, Alane Suhr Published: 2026-06-26 Categories: cs.CV, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2606.28593v1 While recent vision-language models (VLMs) have achieved significant improvements on static visual-to-code tasks such as generating code for webpages, charts, or SVGs, it remains unclear whether they can recover temporal dynamics when motion is present. To this end, we introduce Animation2Code, a benchmark for evaluating temporal visual reasoning via reconstructing executable web animation code from videos. Animation2Code consists of 1,069 web animation videos with diverse visual appearances and motion patterns, paired with corresponding HTML/CSS/JavaScript implementations. We propose two human-aligned metrics, appearance similarity and temporal similarity, which allow us to disentangle visual fidelity from temporal alignment when comparing rendered animations against ground-truth samples. Benchmarking state-of-the-art VLMs on this dataset shows that current VLMs struggle to maintain temporal consistency in reconstruction, even when achieving high appearance similarity, including under finetuning and iterative refinement settings. Code and data are available at https://anya-ji.github.io/animation2code-website . --- ### Search for Truth from Reasoning: A Dynamic Representation Editing Framework for Steering LLM Trajectories Authors: Tianlong Wang, Yuhang Wang, Weibin Liao, Xin Gao, Xinyu Ma, Yang Lin, Yasha Wang, Liantao Ma Published: 2026-06-26 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.28589v2 Current approaches to enhance Large Language Model (LLM) reasoning, such as Chain-of-Thought and "Wait" prompts, primarily encourage models to think more, yet often fail to guide them toward Truth. While Representation Editing (RepE) offers a intrinsic control, its application to dynamic reasoning trajectories remains underexplored. In this work, we bridge this gap by investigating the geometry of truth within unfolding reasoning chains. We uncover three critical insights: (1) Truth is encoded at the sentence level and is entangled with latent reasoning patterns; (2) Effective intervention follows an Uncertainty Principle and a Decay Effect, requiring localization to early, high-entropy forks; (3) Naive steering vectors suffer from noise, risking collateral damage to correct trajectories. Based on these findings, we propose DynaSteer, a dynamic RepE framework. DynaSteer employs pattern clustering to disentangle reasoning manifolds and utilizes Fisher-LDA to project purified truth. By dynamically monitoring lookahead entropy, it selectively steers and rolls back trajectories only when necessary. Comprehensive experimental results on several MATH benchmark verify the effectiveness of DynaSteer, and experiments on out-of-domain coding tasks further confirm its generalization ability. Our code is publicly available at https://github.com/tianlwang/DynaSteer. --- ### SatSplat: Geometrically-Accurate Gaussian Splatting for Satellite Imagery Authors: Shuang Song, Jiyong Kim, Rongjun Qin Published: 2026-06-26 Categories: cs.CV arXiv: https://arxiv.org/abs/2606.28581v1 High-resolution satellite imagery demands 3D reconstruction methods that deliver both speed and geometric accuracy. Recent adaptations of 3D Gaussian Splatting (3DGS) to satellite imagery demonstrate strong efficiency, but reconstruction quality often degrades under diverse illumination across multi-date, high-altitude acquisitions (with small intersection angles), limiting applicability to remote sensing and vision tasks. We present SatSplat, the first framework to adapt 2D Gaussian Splatting (2DGS) to satellite photogrammetry, with online camera adjustment. We approximate satellite cameras with an affine model and learn a minimal delta parameterization for in-splat camera refinement from dense observations. The formulation is implemented with a 2DGS scene representation. To handle time-varying shadows and illumination changes, we integrate geometric shadow mapping and per-camera color correction during training. Across the evaluated DFC2019 and IARPA2016 benchmark sites, SatSplat achieves strong geometric accuracy while significantly outperforming prior 3DGS-based baselines. On our processed DFC2019 benchmark, SatSplat reduces mean absolute error by 11.93% and peak video memory by 31% relative to the previous state of the art. Our approach enables large-scale digital surface modeling with practical computational efficiency. The project page is available at https://gdaosu.github.io/satsplat/. --- ### Surrogate-Gated Generation and Foundation-Model Embeddings for Bayesian Materials Design Authors: Sk Md Ahnaf Akif Alvi, Jan Janssen, Danny Perez, Douglas Allaire, Raymundo Arroyave Published: 2026-06-26 Categories: cond-mat.mtrl-sci, cs.AI arXiv: https://arxiv.org/abs/2606.28578v1 Closed-loop materials discovery iterates between proposing candidate structures and evaluating their properties, and property evaluation dominates the cost. In the generative variant, a learned prior proposes candidate crystals and a property oracle scores them; we ask whether a cheap probabilistic surrogate can triage the generator's output, and what such a surrogate must do well. Across three architecturally distinct pretrained diffusion priors (MatterGen, CrystalFlow, ADiT) and two targets (room-temperature heat capacity and bulk modulus), we insert a Gaussian process acquisition gate between structure generation and the oracle in an RL-steered generative workflow. The gate matches or exceeds ungated fine-tuning of the generative model while capping oracle calls at a fixed per-cycle budget. Budget-matched ablations isolate the mechanism. At an identical four-call budget, ranking-based selection outperforms arbitrary selection, confirming that the gain comes from the surrogate's choice; the gate comes within $\sim$9\% of exhaustive oracle spending at roughly one-fifth of the calls. A density-functional-theory check of the bulk-modulus discoveries confirms the learned oracle to within 2.5\% on average and the surrogate's ranking of the generated structures at Spearman $ρ= 0.94$. A cross-factorial benchmark of surrogate performance spanning mechanical, electronic, and vibrational properties identifies pretrained ORB embeddings with a Gaussian process as the most reliable combination, which we adopt as the building blocks of the proposed workflow. The complete pipeline is released as open-source software. --- ### Correct codes for the wrong reasons? validating LLMs as measurement instruments for theoretical constructs Authors: Manuel Pita Published: 2026-06-26 Categories: cs.CL, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2606.28574v1 When a large language model (LLM) codes a construct in text as a human annotator would, that agreement makes the LLM a reliable coder. Yet reliability leaves construct validity untouched. The instrument may be theory-naive, reaching the code through a correlate that meets none of the demands the construct's theory makes, and no current method tells that apart from genuine measurement. We propose grain calibration as a method that closes the gap. It decomposes a construct into clause-level components, tests each against the text with extractive evidence, and combines the results through an explicit, theory-derived rule. Because the rule is stated rather than lodged in one opaque pass, its structure is evidence about the process rather than the output. It shows which components settled a code, and, when the code is wrong, whether a component was missed or an adjacent construct mistaken for it. Validation shifts from scoring an instrument's outputs against an annotator to showing that the instrument runs on the construct its theory specifies. --- ### Replica Symmetry Breaking and Algorithmic Thresholds in Empirical Risk Minimization under Multi-Index Model Authors: Andrea Montanari, Kangjie Zhou Published: 2026-06-26 Categories: cs.LG, math.ST arXiv: https://arxiv.org/abs/2606.28573v1 Modern machine learning models are trained by optimizing high-dimensional non-convex empirical risk functions. Such cost functions can have a multitude of local optima and yet, gradient-based optimization appears to converge to near-global optima. Within a simple supervised learning setting, we develop a precise picture of which parts of the empirical risk landscape are accessible by polynomial-time algorithms. We are given i.i.d. pairs $\{(\boldsymbol{x}_i,y_i):\; 1 \le i\le n\}$ with $\boldsymbol{x}_i\in \mathbb{R}^d$ standard Gaussian feature vectors, and $y_i\in\mathbb{R}$ response variables that depend on $\boldsymbol{x}_i$ through their projections on an unknown $k$-dimensional subspace. We use empirical risk minimization to learn a model that depends on an $m$-dimensional projection of the data (e.g., an $m$-neurons neural network). We propose an incremental approximate message passing (IAMP) algorithm and precisely characterize the training error it achieves, as well as the relation between test and training error, in the high dimensional asymptotics $n,d\to\infty$, with $n/d\toα\in (0, +\infty)$. Based on earlier work in related models, we expect that the performance achieved by our algorithm is optimal among polynomial-time algorithms. --- ### Geometric Measurements of the Axiom of Choice in Neural Proof Embeddings Authors: Rodrigo Mendoza-Smith Published: 2026-06-26 Categories: cs.LG, cs.AI, cs.LO, math.LO arXiv: https://arxiv.org/abs/2606.28572v1 The axiom of choice has divided the foundations of mathematics for over a century, but the distinction between classical and constructive proofs has remained a philosophical and methodological one. We use Lean 4's kernel-level tracking of axiom dependence to show that the axiom of choice has a measurable geometric correlate in proof space that obeys a one-parameter mixture law and has operational consequences for neural theorem provers. To do this, we partition $471{,}260$ declarations of Mathlib by transitive dependence on the axiom of choice and represent a filtered population of $42{,}355$ traced theorems by their sequences of tactic invocations. We use the constructive proofs in this dataset to train a self-supervised proof encoder and show that when using it to measure classical proofs, three complementary measurements (anomaly score, reconstruction loss, and density-superlevel containment) exhibit a common decline with the proof's distance from the axiom in the dependency graph, from sharp separation at the shallow boundary (AUC $0.847$ at distance $2$) to indistinguishability at distance~$9{+}$. Robustness controls show that the signature survives length, file, author, and topic controls, and replicates under full-source encoders trained on normalised proof source. Operationally, we show that on an evaluation sample of $251$ Mathlib theorems, Lean's \texttt{aesop} tactic solves constructive theorems at $13\times$ the rate of classical ones, and a neural-guided hybrid using the ReProver tactic generator compresses the gap to $5\times$. The geometric anomaly score predicts \texttt{aesop} failure beyond proof length, providing an operational link between the geometric signature and prover performance. --- ### Digitizing Coaching Intelligence: An Agentic Framework for Holistic Athlete Profiling using VLM and RAG Authors: Deep Ghosal, Ishani Sen, Wazib Ansar, Amlan Chakrabarti Published: 2026-06-26 Categories: cs.CV, cs.AI, cs.MA arXiv: https://arxiv.org/abs/2606.28570v1 Athlete assessment is a critical process for tracking physical progress and identifying elite talent. However, during mass recruitment drives, traditional methods rely on manual observation, which is inherently subjective and unscalable, or basic computer vision (CV) systems limited to quantitative repetition counting. These standard approaches lack the "coaching intelligence" required to evaluate qualitative physiological markers such as form degradation, spinal articulation, and fatigue. This paper presents a novel, LLM-based hybrid agentic framework for automated, holistic athlete profiling that strictly aligns with the Sports Authority of India (SAI) assessment protocols. Orchestrated via LangGraph, our dual-pipeline architecture synthesizes the geometric precision of CV (MediaPipe) for kinematic tracking with the semantic reasoning of Vision-Language Models (Llama-4-scout). To overcome the latency and token constraints associated with multimodal video processing, we introduce a 3 X 3 "Smart Grid" temporal chunking strategy, reducing computational overhead by over 88% while preserving critical temporal continuity. To ensure data integrity and mitigate hallucination, the framework pioneers an autonomous "LLM-as-a-Judge" self-correction loop that cross-references quantitative and qualitative metrics before persistence. Finally, we implement a dual-persistence Retrieval-Augmented Generation (RAG) pipeline utilizing a vector search engine (ChromaDB). This enables coaches to bypass rigid SQL databases and perform complex semantic queries (e.g., "Identify athletes with high endurance but poor core rigidity") using natural language. Experimental results demonstrate that this multi-agent approach significantly bridges the gap between raw biometric tracking and actionable coaching insights, offering a scalable, objective solution for national talent identification. --- ### KM-Speaker: Keypoint-Based Style Control for High-Quality Speech-Driven 3D Facial Animation and Dialogue Localization Authors: Arthur Josi, Emeline Got, Abdallah Dib, Luiz Gustavo Hafemann, Rafael M. O. Cruz Published: 2026-06-26 Categories: cs.CV, cs.GR, cs.LG arXiv: https://arxiv.org/abs/2606.28568v1 Speech-driven 3D facial animation methods face significant challenges in simultaneously achieving high-fidelity motion and precise artistic control at production quality. Existing controllable models typically learn global style control by relying on large-scale, low-quality \emph{in-the-wild} datasets that compromise overall animation realism. Furthermore, these frameworks often lack the fine-grained temporal precision required for demanding tasks such as dialogue localization (e.g., dubbing), where matching specific facial expressions is as critical as lip synchronization. We present KM-Speaker (Keypoint-Matching Speaker), a novel keypoint-conditioned flow-based generative framework that provides both global style guidance and frame-level temporal control from reference performances. We propose a disentanglement strategy that separates audio-driven lip motion from keypoint-driven upper-face dynamics, together with a global style context preservation mechanism to ensure coherent full-face expressiveness. KM-Speaker advances example-based 3D facial animation by achieving high-fidelity motion and flexible controllability in a data-constrained setting, consistently outperforming state-of-the-art methods in lip-sync accuracy, style adherence, and expressive temporal control. --- ### KernelSight-LM: A Kernel-Level LLM Inference Simulator Authors: Xiteng Yao, Taeho Kim, Hengzhi Pei, Xinle Liu, Kyle Ulrich, Leonard Lausen, Ashish Khetan, Xiang Song, George Karypis, Martin Herbordt Published: 2026-06-26 Categories: cs.PF, cs.AI, cs.AR arXiv: https://arxiv.org/abs/2606.28565v2 As large language models (LLMs) move into production serving, practitioners must rapidly evaluate inference performance across diverse hardware, models, and serving parameters to meet cost and latency targets. However, the end-to-end behavior of LLMs couples serving-layer policies with low-level GPU kernel execution and rapidly evolving architectures, forcing slow, deployment-specific benchmarking that is hard to generalize. We present KernelSight-LM, a fine-grained inference simulator that models token-level execution and produces kernel-level latency breakdowns. It decomposes each serving step into a roofline kernel model with a learned efficiency term, a communication model, and a host-overhead model, composed through a discrete-event scheduler that also captures mechanisms like prefix caching and continuous batching. KernelSight-LM offers two prediction tiers that trade target-GPU data for accuracy. The cross-generation tier uses no target-GPU measurements, only hardware specifications and kernel microbenchmarks from previously profiled GPUs, and predicts per-kernel latency on an unseen GPU generation to 12.1% error, a 1.8x improvement over the roofline baseline (22.0%). A second target-measured tier adds one model-agnostic kernel-microbenchmark sweep on the target GPU, sharpening per-kernel error to 3.8%, a 7.3x improvement over a comparable baseline (27.7%). Both tiers require far less target-GPU data than the prior systems they extend. In our simulator, these predictions yield end-to-end median (p50) errors across six model families of 15.4%, 12.8%, and 3.0% (TTFT, TPOT, throughput) in the cross-generation tier and 14.3%, 6.2%, and 2.7% in the target-measured tier, matching dedicated profiling tools while collecting far less on-device data. Beyond prediction, its kernel-level bottleneck breakdowns support hardware/software co-design and capacity planning. --- ### SEAD: Competence-Aware On-Policy Distillation via Entropy-Guided Supervision Authors: Chia-Hsuan Lee, Zelei Cheng, Yu Wang, Renkun Ni, Sambit Sahu, Shi-Xiong Zhang, William Campbell Published: 2026-06-26 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.28562v1 On-policy distillation (OPD) has a property absent in offline distillation and RL: teacher supervision quality depends on student competence. Incoherent rollouts yield noisy gradients; already-mastered tokens yield redundant ones. This creates waste at three scales (tokens, training phases, and prompts) yet existing methods supervise uniformly. We introduce SEAD, which uses entropy as a unified probe of this competence-dependent degradation at three scales: (1) joint teacher-student entropy partitions tokens into zones receiving tailored divergences or zero gradient (approx. 50% skipped); (2) a cosine schedule anneals from forward to reverse KL as competence grows; (3) a competence-gated curriculum introduces prompts easy-to-hard. These components are symbiotically necessary: token selection requires coherent rollouts (curriculum), annealing requires monotonic improvement (also curriculum). On OLMo-3 (7B to 32B), SEAD achieves +4.8 avg accuracy over vanilla OPD across six math benchmarks, with ablations confirming super-additive interactions. --- ### Depth-Staggered Fibonacci Spacing for Sparse Attention: Static Schedules Beat Learned Dilation and Extrapolate Where Dense Attention Fails Authors: Chad A. Capps Published: 2026-06-26 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.28560v1 We study sparse self-attention in which each query attends to a dense local window plus a set of Fibonacci-spaced offsets, with a per-layer scalar alpha that compresses or expands the spacing. Across 21 language models trained under one matched recipe (60M parameters, 512 hidden, 16 layers, 426M tokens), we compare four ways of setting alpha across depth: fixed, per-layer learned, a static linear stagger, and a coprime (anti-gridding) reassignment of that stagger, together with a reach-matched power-of-2 control. Three results stand out. First, a static per-layer stagger improves perplexity over both fixed and learned alpha, and the gain is base-agnostic: applying the same stagger to a power-of-2 base lifts it above fixed Fibonacci and to parity with learned Fibonacci attention. Second, learning per layer is inert: it does not beat the static schedule and costs roughly five times the inference latency. Third, and most consequential, all sparse variants extrapolate to four times their training length with little or no degradation, whereas a recipe-matched dense baseline collapses (perplexity rises by 201% at 4x length); we attribute this to fixed-offset attention only ever querying relative positions seen during training. We also report two honest negatives: at training length the best sparse model has about 26% higher perplexity than the dense baseline, and the staggering gain is uniform across context positions rather than concentrated at long range. --- ### IMCBench: A benchmark for multimodal LLMs in Image-grounded Medical Conversations Authors: Maria Xenochristou, Ashutosh Joshi, Korosh Vatanparvar, Mohammad Abuzar Hashemi, Prasad Kasu, Deepak Bansal, Anchal Nema, Nivedita Wadhwa, Prashams S Jain, Rebecca Abraham, Will Kimbrough, Dilek Hakkani-Tur et al. Published: 2026-06-26 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.28556v1 Recent advances in large language models and vision-language models have enabled reasoning over multimodal data, offering opportunities for clinical applications such as decision support and triaging. However, existing medical AI benchmarks are fragmented: some support multi-turn dialogues but lack images, while others provide multimodal inputs but focus on single-turn QA tasks. To address this gap, we introduce IMCBench, an image-grounded, multi-turn medical conversation benchmark that pairs real, publicly available clinical images with synthetic patient profiles to simulate realistic patient-clinician interactions. Each conversation is evaluated across three clinical dimensions: safety, accuracy, and appropriate use of uncertainty in diagnosis. We benchmark eight multimodal frontier models across four model families (Claude, GPT, Nova, and Llama), scoring each on a 1-5 scale using LLM-as-Jury scoring calibrated against expert clinician annotations. Our results show that Claude Opus 4.6 achieves the highest overall score (3.61), followed by Claude Sonnet 4.6 (3.30) and GPT-5.2 (3.29), though no model dominates all dimensions and safety degrades for both malignant and rare conditions ($Δ$ = -0.27 each). Ablation studies further reveal that both visual input and EHR context contribute to safe guidance (safety drops of 0.18 and 0.23 on average when each is removed), with stronger models leveraging visual features more effectively. Together, these findings demonstrate that accurate clinical description does not guarantee safe patient guidance, motivating the need for multi-dimensional evaluation frameworks in medical AI. --- ### Improving Coherence in Hierarchical Time Series Forecasting using Structured Temporal Fusion Authors: Ruchi Pakhle Published: 2026-06-26 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.28553v1 In many real-world applications, such as retail sales, energy usage, and supply chain planning, forecasting is performed across hierarchical structures. These structures often represent aggregations (e.g., products to categories to regions), where forecasts must not only be accurate but also coherent, meaning that lower-level predictions sum correctly to higher-level forecasts. Traditional statistical methods, such as Bottom-Up and MinT, enforce coherence through post-processing but fail to model complex nonlinear temporal dependencies and covariate interactions. We propose Hierarchical Temporal Fusion (HTF), a novel extension of the Temporal Fusion Transformer (TFT) that integrates structured hierarchical embeddings with a coherence-aware loss function to ensure consistent forecasts across all levels of a hierarchy. Rather than applying reconciliation after forecasting, HTF embeds coherence directly into the training objective. The coherence loss penalizes the difference between aggregated child forecasts and their corresponding parent forecasts during training, enabling the model to learn both temporal dynamics and structural consistency simultaneously. We evaluate HTF on two publicly available benchmark datasets: the M5 Walmart forecasting dataset and a publicly available hierarchical energy consumption dataset. Results demonstrate that HTF substantially reduces forecast incoherence while improving forecasting accuracy compared with classical reconciliation methods and deep learning baselines. In addition, attention visualization and embedding analysis provide insight into how temporal and structural information contribute to hierarchical forecasting performance. --- ### DataComp-VLM: Improved Open Datasets for Vision-Language Models Authors: Matteo Farina, Vishaal Udandarao, Thao Nguyen, Selim Kuzucu, Maximilian Böther, Andreas Hochlehnert, Adhiraj Ghosh, Marianna Nezhurina, Karsten Roth, Joschka Struber, Yuhui Zhang, Sebastian Dziadzio et al. Published: 2026-06-26 Categories: cs.CV, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.28551v3 Building performant Vision-Language Models (VLMs) requires carefully curating large-scale training datasets, yet the community lacks systematic benchmarks for evaluating such curation strategies. We introduce DataComp for VLMs (DCVLM), a benchmark for controlled data-centric experiments to improve VLM training. As part of DCVLM, we collect 160 datasets spanning four data types -- image-caption pairs, multimodal interleaved documents, text-only, and instruction-tuning data -- into a corpus of 6T multimodal tokens. DCVLM allows participants to test curation strategies (filtering, mixing, formatting, sampling) across 1B-8B models and 6.25B-200B token budgets. Models are then evaluated on a carefully selected suite of up to 52 downstream benchmarks across 9 domains. We conduct extensive experiments on DCVLM and find that data mixing, not filtering, is key to a high-quality training dataset: instruction-heavy mixtures scale better than caption-heavy ones, with gains widening at larger scales. The resulting dataset, DCVLM-Baseline, enables training an 8B VLM to 63.6% accuracy on our 33-task core suite with 200B training tokens. Compared to FineVision, the state-of-the-art open VLM training dataset, this represents an improvement of +5.4pp. DCVLM and all accompanying artifacts will be made publicly available at https://www.datacomp.ai/dcvlm/. --- ### DataComp-VLM: Improved Open Datasets for Vision-Language Models Authors: Matteo Farina, Vishaal Udandarao, Thao Nguyen, Selim Kuzucu, Maximilian Böther, Andreas Hochlehnert, Adhiraj Ghosh, Marianna Nezhurina, Karsten Roth, Joschka Struber, Yuhui Zhang, Sebastian Dziadzio et al. Published: 2026-06-26 Categories: cs.CV, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.28551v2 Building performant Vision-Language Models (VLMs) requires carefully curating large-scale training datasets, yet the community lacks systematic benchmarks for evaluating such curation strategies. We introduce DataComp for VLMs (DCVLM), a benchmark for controlled data-centric experiments to improve VLM training. As part of DCVLM, we collect 160 datasets spanning four data types -- image-caption pairs, multimodal interleaved documents, text-only, and instruction-tuning data -- into a corpus of 6T multimodal tokens. DCVLM allows participants to test curation strategies (filtering, mixing, formatting, sampling) across 1B-8B models and 6.25B-200B token budgets. Models are then evaluated on a carefully selected suite of up to 52 downstream benchmarks across 9 domains. We conduct extensive experiments on DCVLM and find that data mixing, not filtering, is key to a high-quality training dataset: instruction-heavy mixtures scale better than caption-heavy ones, with gains widening at larger scales. The resulting dataset, DCVLM-Baseline, enables training an 8B VLM to 63.6% accuracy on our 33-task core suite with 200B training tokens. Compared to FineVision, the state-of-the-art open VLM training dataset, this represents an improvement of +5.4pp. DCVLM and all accompanying artifacts will be made publicly available at https://www.datacomp.ai/dcvlm/. --- ### Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution Authors: Kevin Der, Harish Kamath, Ben Thompson Published: 2026-06-26 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.28548v1 Sparse autoencoders (SAEs) have become a useful tool for extracting interpretable features in language models. However, standard SAE architectures operate on individual token activations, meaning that the number of active features scales linearly with context length, and studying long model transcripts becomes difficult. We introduce turn-averaged SAEs, which represent a single Human or Assistant turn with a fixed number of features by learning to reconstruct the average model activation across the turn. We find that turn-averaged features describe a single turn's high-level characteristics more completely than per-token features when judged by an LLM. We also demonstrate that turn-averaged SAEs greatly simplify common downstream uses of SAEs like attribution graphs. Broadly, turn-averaged SAEs make interpretability techniques practical at long context lengths. --- ### NIVA: A Multimodal Foundation Model for Actionable Earth System Intelligence Authors: Anisha Pal, Aodhan Sweeney, Kyle Heyblom, Kalai Ramea Published: 2026-06-26 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.28546v1 Recent advances in AI-driven weather and climate modeling have improved forecast skill while reducing computational cost. However, existing data-driven approaches are limited in their ability to model coupled Earth system dynamics, which is required for extending predictability beyond the ~2-week horizon. To address this, we introduce NIVA, a multimodal foundation model designed to learn unified representations across Earth system components. While the full framework targets atmosphere, ocean, ice, and land interactions, we focus here on a two-modality setting (ocean and atmosphere) as a controlled proof of concept to evaluate whether foundation models can learn coupled dynamics. Trained on large-scale Earth system simulations, NIVA learns physically meaningful cross-modal structure, providing a foundation for subseasonal-to-seasonal prediction. As initial validation, we show that NIVA captures key modes of climate variability through accurate prediction of major climate indices. --- ### Who Plays Which Role When? Communication Role Dynamics for Peer Recognition and Team Performance Prediction Authors: Yifan Song, Wenxuan Wendy Shi, Brian P Bailey, Tal August Published: 2026-06-26 Categories: cs.CY, cs.CL arXiv: https://arxiv.org/abs/2606.28544v1 Team roles offer an interpretable lens on collaboration, yet computational studies of roles often rely on domain-specific personas or data-driven clustering rather than theory-grounded taxonomies. We operationalize a taxonomy of eight communication roles grounded in education literature and annotate a corpus of 6,307 Slack messages from 55 students across 18 teams in a semester-long computer science course project. We evaluate whether LLMs can approximate expert labels, enabling scalable, taxonomy-driven role annotation. Using these role labels, we characterize role dynamics over teams' lifecycles, finding that different roles peak at different moments and that students enact a more diverse set of roles as projects progress. To evaluate the utility of our role constructs, we use them to predict peer recognition, outperforming lexical, conversational, and LLM-prompting baselines. To assess generalizability beyond the educational context, we apply the same role constructs to a public dataset (DeliData) to predict team performance improvement after deliberation, again exceeding prior performance. --- ### Comparing Scalar Objective Functions for Multi-Criteria Engineering Optimization Authors: Olaf Frommann Published: 2026-06-26 Categories: math.OC, cs.CE, cs.NE arXiv: https://arxiv.org/abs/2606.28541v1 Scalar objective functions are required when a multi-criteria optimization problem must yield a single preferred design rather than only a Pareto set. The choice of scalarization influences which compromise is selected, how preference parameters are interpreted, and whether non-supported Pareto regions can be reached. This paper compares four formulations for normalized bi-criteria minimization: weighted sums, achievement scalarizing functions, desirability functions, and a fuzzy-logic-based formulation. Two analytically defined Pareto fronts, one convex and one concave, isolate the effect of the objective formulation from numerical optimizer behavior. The comparison focuses on reachable Pareto regions, parameter-induced selection density, compensation between criteria, sensitivity, and interpretability. Results show that weighted sums are simple but structurally limited on concave fronts, while achievement, desirability, and fuzzy formulations reach interior non-supported regions through different mechanisms. Desirability functions introduce nonlinear single-criterion preference mappings, whereas fuzzy rules express nonseparable and reference-dependent engineering preferences. --- ### Legal Domain Adaptation of Modern BERT Models Authors: Dominik Stammbach, Peter Henderson Published: 2026-06-26 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.28538v1 We investigate domain adaptation of modern BERT models in the legal domain. We further pre-train ModernBERT on all US court opinions using the masked language modeling objective. Although ModernBERT has been trained on roughly 500x more data than original BERT, we still find that this model benefits from further pre-training and domain adaptation in the legal domain: we report significant improvements compared to vanilla ModernBERT on all datasets connected to US court opinions. We find gains similar to those reported in early work on domain adaptation of BERT-like models. However, from scratch pre-training does not match the performance of further pre-training an existing ModernBERT checkpoint in our experiments. The resulting models are capable of processing sequences up to 8,192 tokens, and can be used to compute meaningful embeddings of legal passages, or could quickly rerank hundreds of legal passages for a given search query. We release all model checkpoints publicly. --- ### MammoFlow: Multiview Mammogram Synthesis with Anatomically Consistent Flow Matching Authors: Yuexi Du, Leya Barrientos, Laura Sheiman, John Lewin, Hemant D. Tagare, Nicha C. Dvornek Published: 2026-06-26 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2606.28537v1 Multiview mammography relies on paired craniocaudal (CC) and mediolateral oblique (MLO) views to provide complementary projections of a 3D breast volume, enabling precise anomaly localization. However, acquiring high-quality, balanced datasets remains challenging for deep learning applications. We propose a novel method to synthesize multiview mammograms by leveraging the inherent geometric relationship between CC and MLO views. To enforce an implicit 3D consistency prior during generation, we develop an alignment module that searches a 2D affine transformation subspace to establish optimal anatomical correspondence. Leveraging this alignment, we introduce a pixel-space self-consistency loss based on the Earth Mover's Distance (EMD) between the 1D anteroposterior (AP) axis tissue distributions of the generated images. Integrated into a pretrained flow matching model, MammoFlow forces synthesized pairs to share physically plausible tissue distributions from the chest wall to the nipple. To our knowledge, this is the first work to guide multiview mammogram generation using implicit geometric tissue correspondence. Our method demonstrates superior image quality, passes expert radiologist evaluation, and generates physically consistent pairs that improve downstream classification AUC by 5%. Code is available at https://github.com/XYPB/MammoFlow --- ### CMSL: Constructive Multi-Sequence Learning for Recommendation Systems Authors: Zikun Cui, Renzhi Wu, Junjie Yang, Li Sheng, Jijie Wei, Linfeng Liu, Tai Guo, Tao Jia, Xiaodong Wang, Hong Li, Li Yu, Sri Reddy et al. Published: 2026-06-26 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2606.28533v2 Sequence learning has emerged as the promising paradigm in recommendation systems, surpassing traditional Deep Learning Recommendation Models (DLRM) by capturing the temporal nuances of user behavior. However, current state-of-the-art architectures operate under a limiting analogy: they treat user history as a monolithic chronological sequence like a sentence in a Large Language Model (LLM). We observe a fundamental divergence between natural language and recommendation data: unlike the linear, logical flow of text, user history is inherently multi-faceted. A user's journey is a fragmented reflection of diverse interests, resulting in much weaker coherence between items than is found in LLM training data. This lack of structural unity leads to context pollution. In single-sequence modeling, unrelated behaviors compete for the same attention budget. This "noisy" signal dilutes the model's focus, effectively capping its ability to discern high-intent patterns from background activity. To address this, we propose Constructive Multi-Sequence Learning (CMSL), a paradigm shift from passive sequence ingestion to active "context engineering" that constructs multiple coherent sequences in latent space. CMSL leverages a learnable Sequence Construction Module to disentangle user history into "pure" thematic strands, followed by a linear attention mechanism to efficiently model these strands at scale. CMSL has been deployed across ranking and retrieval tasks and across four major surfaces at Meta. --- ### A Good Talk Does not Look Like a Summary, It Teaches You! Measuring Takeaways from Paper-to-Video Talks Authors: Ishani Mondal, Aparna Garimella, Ananya Sai, Pannaga Shivaswamy, Jordan Boyd-Graber Published: 2026-06-26 Categories: cs.MM, cs.CL arXiv: https://arxiv.org/abs/2606.28531v1 Automatically generated videos from scientific papers are increasingly used for education and research dissemination. However, existing evaluation metrics mainly measure visual quality or whether key points from the paper appear in the video without assessing whether the video actually helps viewers understand the ideas. We introduce EffectivePresentationScorer, a framework for evaluating the instructional quality of scientific presentation videos. It checks whether a video explains the main ideas clearly, introduces needed background concepts, and connects technical details to the main contribution of the paper. When we apply EffectivePresentationScorer to the existing paper-to-video generation systems, we find that generated videos mention the correct topics and follow the structure of the paper but fail to explain prerequisite concepts or clarify why the method works. These failures are often ignored by existing video evaluation metrics, which focus on content presence rather than explanatory quality. --- ### The Speedup Paradox: Rethinking Inference Speed-Quality Trade-off in Embodied Tasks Authors: Yujin Wang, Junli Chen, Yixuan Li, Shunan Dong, Huazhong Yang, Yongpan Liu, Hongyang Jia Published: 2026-06-26 Categories: cs.RO, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2606.28529v2 Embodied foundation models have recently been widely used to improve robot generalization and task success rates. Previous works apply lossy efficient-inference techniques such as quantization, pruning, and asynchronous inference, accepting small action quality degradation in exchange for lower per-step computation cost and inter-action latency. However, unlike traditional static ML tasks, embodied tasks involve repeated interaction with the environment, and task-level performance is determined not only by per-step cost, but also by closed-loop effects unique to embodied execution, which remain insufficiently characterized in current efficient-inference studies. In this work, we propose TISED (\underline{T}ask-level \underline{I}nference \underline{S}peedup \underline{E}ffect \underline{D}ecomposition), an analytical framework that unifies diverse lossy inference optimization techniques and decomposes their effects on static and dynamic tasks, and uncovers some paradoxical effects on task-level performance: (1) on \textit{static tasks}, optimization sometimes can lengthen end-to-end per-task completion time even as per-step latency drops; (2) on \textit{dynamic tasks}, moderate lossy optimization can raise task success rate even above the baseline; and (3) the monotonicity and sweet-spot location of both effects can shift with hardware configuration. Together, our findings provide a new perspective on adapting inference optimization techniques to embodied tasks. --- ### A French OSCE Dialogue Dataset and Controllable Virtual Patient System for Clinical Training Authors: Doria Bonzi, Tom Bourgeade, Fabrice Lefèvre, Irina Illina Published: 2026-06-26 Categories: cs.CL, cs.HC arXiv: https://arxiv.org/abs/2606.28526v1 The clinical and communication skills of medical students are commonly assessed through Objective Structured Clinical Examinations (OSCEs), which consist of brief scenario-driven simulations of doctor-patient interactions. However, training is often limited by the low availability of human standardized patients, motivating the development of realistic virtual patients (VPs). To address this gap, we introduce a French OSCE dialogue dataset comprising 240 student-patient training interactions. We build upon it a controllable LLM-based pipeline to generate synthetic OSCE dialogues. The pipeline integrates modular components, such as retrieval-based grounding and a reflection loop, to ensure patient fidelity, coherence, and realism. Additionally, we propose a multi-level evaluation framework assessing patient simulation quality, student performance, and linguistic quality, using an LLM-as-a-Judge approach. Experiments suggest that controllability modules generally improve patient fidelity and student evaluation consistency. Finally, we implement an interactive prototype in which students can practice with a VP and receive automatic feedback. --- ### A Gravitational Interpretation of Safety Reversion under Fine-Tuning Authors: Samuele Poppi, Nils Lukas Published: 2026-06-26 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.28525v2 Safety alignment in large language models can degrade during post-training even when neither the data nor the objective is intentionally adversarial. Alignment rebound and reverse dynamics suggest that this degradation may reactivate behavior suppressed during safety alignment. Building on these ideas, we hypothesize that ordinary non-adversarial post-training follows a reversion direction: the activation-space displacement from the safety-aligned model toward a more permissive, earlier helpful-only state. We see that for Llama, every tested trajectory across references, tasks, and seeds exceeds a matched empirical null, while at aligned Llama and Qwen checkpoints, a vocabulary readout shows that the direction locally favors task-engaging over fixed refusal-like openings. Its geometric expression is behaviorally informative: as post-training proceeds, alignment with the direction and harmfulness increase together, yielding a strong descriptive correlation (Spearman r=0.958). To move beyond correlation, we test causal relevance during adaptation using objectives constructed from this coordinate. Across all tested Llama, Qwen, and Gemma settings from 3B to 14B, an optimizer-matched objective opposing positive motion reduces geometric alignment and harmfulness relative to ordinary fine-tuning, whereas a separately stabilized objective reinforcing that motion increases both. Every model and scale exhibits the same mean block-baseline-push ordering, showing that the causal relevance of the reversion direction is not tied to one architecture or model size. Finally, we show that a standard safety-rehearsal objective, built without access to the direction, independently opposes it and cuts cumulative reversion by about 30% in Llama and Qwen. --- ### A Gravitational Interpretation of Fine-Tuning Reversion Authors: Samuele Poppi, Nils Lukas Published: 2026-06-26 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.28525v1 Fine-tuning on harmless data can partially undo behaviors acquired earlier in training. Safety can erode under benign post-alignment updates, unlearned capabilities can re-emerge, latent traits can transfer through apparently unrelated supervision, and related post-alignment fragility appears in other generative settings. We argue these phenomena are usefully viewed through a common training-history lens. Our hypothesis is geometric: large early training phases create dominant behavioral manifolds, while later alignment or specialization phases are shallower displacements from them. Subsequent fine-tuning can therefore inherit a persistent reversion component pointing back toward a witness of the dominant manifold. We call this the gravitational interpretation of fine-tuning reversion. Across our main settings, representational drift rapidly acquires a component along a history-defined reversion direction (v_rev). In our main track, alignment with v_rev rises from cos = 0.429 +/- 0.052 after the first update to 0.647 +/- 0.021 by step 20. Across 24 run-step pairs, every observed alignment exceeds the p99 of an isotropic activation-space null. We demonstrate that selectively blocking motion along v_rev changes the final alignment at T=100 from 0.648 +/- 0.009 to -0.211 +/- 0.021 and reduces harmfulness from 19.0% +/- 4.0% to 8.5% +/- 1.5% with little task cost. These results support v_rev as a causally relevant mediator of early post-alignment reversion in our setup. Importantly, we do not claim that v_rev is the unique safety direction, nor that the dominant manifold is directly observed; rather, we identify a robust, history-defined direction that explains and partially controls early reversion dynamics. --- ### Developmental Trajectories of Situation Modeling and Mentalizing in Transformer Language Models Authors: Pamela D. Rivière, Cameron Jones, Sean Trott Published: 2026-06-26 Categories: cs.CL arXiv: https://arxiv.org/abs/2606.28524v1 Recent work suggests that Large Language Models (LLMs) are sensitive to the belief states of agents described by text, as measured by the false belief task (FBT), yet persistent concerns of construct validity remain. We adopt a **developmental perspective**, tracing the pattern of mental state reasoning behavior -- and likely **preconditions** for this behavior -- across multiple training stages in the Olmo2 and Pythia language model suites. We find that above-chance FBT performance depends both on model size and sufficient training volume, emerges relatively late in pretraining, and is most improved by post-training interventions (SFT, DPO) in the condition most diagnostic of mentalizing (False Belief, Implicit). However, FBT performance is fragile: consistent with past work, the use of non-factive verbs (e.g., thinks) increases false belief attributions even in the True Belief condition. To contextualize these findings, we track the emergence of **situation modeling**: the ability to report on basic factual properties of a described scene. Situation modeling accuracy generally precedes and exceeds FBT accuracy, yet situational representations also prove surprisingly incoherent in certain respects: when asked about the knowledge states of the Antagonist agent -- who always knows the item's true location -- Olmo2 13b is consistently influenced both by the Target agent's knowledge state and the presence of non-factive verbs. Together, these results suggest that larger, sufficiently trained models build partially coherent situation models in a developmentally appropriate sequence, yet display surprising fragility -- highlighting the value of developmental and stress-testing approaches for evaluating LLM capabilities. --- ### Detecting Clinical Hallucinations in LVLMs via Counterfactual Visual Grounding Uncertainty Authors: Xiao Song, Haonan Qin, Zhaoxu Zhang, Jiong Zhang, Yuqi Fang, Caifeng Shan Published: 2026-06-26 Categories: cs.CV, cs.CL arXiv: https://arxiv.org/abs/2606.28520v1 Large vision-language models (LVLMs) are increasingly used for clinical image understanding, yet they remain vulnerable to \emph{hallucinations}--producing textual findings or attributes not supported by the image. We present a vision-traceable hallucination detection framework that audits arbitrary LVLM responses via visual evidence grounding, requiring neither modification nor internal access to the hidden states of LVLMs. Given an LVLM response, we extract visually verifiable entities and use a medical-domain-adapted Qwen-VL grounding verifier to localize each entity on the input image. To enhance the robustness of our detection method, we introduce a counterfactual entity perturbation method and estimate visual evidence uncertainty by contrasting factual and counterfactual grounding results. Specifically, we compute an entity-level uncertainty score from the positive confidence, counterfactual confidence, and their grounding overlap for binary hallucination decision-making. Experiments on multiple medical imaging modalities and LVLM backbones demonstrate that our method consistently improves hallucination detection performance over recent baselines, while providing interpretable localization evidence and strong cross-model transferability. Code and dataset are available at https://github.com/Agentic-CliniAI/CounterVHD. --- ### A Trainable-by-Parts Operator Learning Framework: Bridging DeepONet and Karhunen-Loeve Expansions for Large-Scale Applications Authors: Christian Munoz, Alexandre Tartakovsky Published: 2026-06-26 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.28519v1 Training operator-learning models for large-scale problems governed by partial differential equations (PDEs) is challenging due to the curse of dimensionality, memory constraints, and limited training data. These challenges arise in many scientific and engineering applications, including subsurface flow, climate modeling, and geological carbon storage (GCS). In this work, we propose a scalable operator-learning framework based on the Karhunen-Loeve Deep Neural Network (KL-DNN) and demonstrate its performance for modeling GCS. The model is trained on a dataset comprising 100 samples of large-scale simulations in a three-dimensional domain with 1.7 million cells and 50 time steps. The KL-DNN method constructs latent spaces using low-rank singular value decomposition of static properties and a nested Karhunen-Loeve expansion for dynamic pressure fields, enabling full-resolution predictions without subsampling or spatial coarsening. The KL-DNN model achieves an average root mean square error (RMSE) of 1.1 psi for pressure (0.04% relative error with respect to the average pressure in the domain) and RMSE of 0.0146 for CO2 saturation (5% relative error with respect to the average saturation inside the plume). The model requires 20 minutes of training on a single GPU, representing a 19% reduction in the pressure errors, 7% reduction in the saturation error, and a two-order-of-magnitude speedup compared to DeepONet trained on the same dataset. These results, along with inference time of less than one minute, establish the proposed model as a practical and accurate solution for large-scale PDE problems, enabling rapid uncertainty quantification, history matching, and real-time decision support. --- ### CLEAR-MoE: Shared-Basis Expert Extraction from Frozen Vision Transformers via Calibration-Driven Layer Selection Authors: Md Irtiza Hossain, Humaira Ayesha, Junaid Ahmed Sifat Published: 2026-06-26 Categories: cs.CV, cs.DC arXiv: https://arxiv.org/abs/2606.28516v1 We present CLEAR-MoE, a four-phase post-training pipeline that converts a frozen pretrained Vision Transformer (ViT) into a sparse Mixture-of-Experts (MoE) model without updating backbone weights. The pipeline (i) scores feed-forward network (FFN) layers by sparsity, clusterability, and output sensitivity; (ii) decomposes selected layers into a shared low-rank SVD basis and per-cluster residual experts using k-means clustering; (iii) trains lightweight routers supervised by cluster labels; and (iv) dispatches tokens through pluggable CUDA backends. On Imagenette with DeiT-Small, CLEAR-MoE retains 99.9% of the dense model's accuracy (86.70 +/- 0.02% versus 86.73%). Extensive ablation studies reveal a consistent empirical finding: the shared SVD basis is the primary factor responsible for preserving accuracy. Random routing, learned routing, and three different router architectures produce nearly identical performance, with accuracy varying by at most 0.06 percentage points (86.62%-86.68%). Accuracy also remains stable across different SVD ranks, expert counts (2-8), calibration set sizes (50-500), and random seeds. This behavior generalizes across five ViT backbones (DeiT-Tiny, DeiT-Small, DeiT-Base, ViT-Small, and ViT-Base), covering models from 5.7M to 86.6M parameters, with accuracy differences <= 0.10 percentage points from their dense counterparts. On a GTX 960 GPU, routing and scatter-gather overhead make the CLEAR-MoE FFN 1.3-1.7x slower than the dense implementation. A dispatch microbenchmark further shows that routing is an order of magnitude more memory-bound than expert matrix multiplications, identifying fused dispatch kernels as a promising direction for future optimization. --- ### GPTNT: Benchmarking Real-Time Collaboration Between Multimodal Agents on Keep Talking And Nobody Explodes Authors: Amit Parekh, Sabrina McCallum, Kareem Al-Hasan, Malvina Nikandrou, Alessandro Suglia, Ioannis Konstas Published: 2026-06-26 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2606.28514v2 Multimodal models are increasingly deployed to solve tasks collaboratively with humans or other artificial agents. While existing benchmarks show that they possess the fundamental capabilities, the various conditions that coincide when collaborating---time pressure, information asymmetry, and imperfect communication---have traditionally been studied in isolation. To address this gap, we introduce GPTNT, a benchmark built on the cooperative video game Keep Talking and Nobody Explodes, in which two agents must coordinate to defuse procedurally generated bomb puzzles against a live countdown. One agent has access to the bomb but not the instructions for defusing it; the other holds the instructions but cannot see or manipulate the bomb. Neither agent can succeed alone: the task requires contributions from both, and is solvable only through effective, efficient communication. We remove turn-taking proxies or simplifications, instead requiring agents to act asynchronously and communicate in real time. GPTNT is designed to expose how models collaborate versus how they perform alone: the instruction manual, the partner, or both, can optionally be withheld to surface what a model has memorised versus what it derives in the moment. We demonstrate that GPTNT poses a considerable challenge to the state-of-the-art: not one of the closed- and open-source models we test defuses a single bomb in real time, a bar that human players clear. In a range of controlled experiments, we explore where capabilities break down, identifying critical weaknesses in state tracking, efficient acting within the time budget, handling ambiguity, and error recovery. Since it runs on the real game, GPTNT benefits from procedural generation and inherits a living modding community: as models improve, the benchmark can be evolved to remain challenging, rather than being solved once and retired. --- ### GPTNT: Benchmarking Real-Time Collaboration Between Multimodal Agents on Keep Talking And Nobody Explodes Authors: Amit Parekh, Sabrina McCallum, Kareem Al-Hasan, Malvina Nikandrou, Alessandro Suglia, Ioannis Konstas Published: 2026-06-26 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2606.28514v1 Multimodal models are increasingly deployed to solve tasks collaboratively with humans or other artificial agents. Existing benchmarks show that these models possess many of the required component capabilities, but the conditions that coincide in collaboration, including time pressure, information asymmetry, and imperfect communication, are usually studied in isolation. We introduce GPTNT, a benchmark built on the cooperative video game Keep Talking and Nobody Explodes, in which two agents must coordinate to defuse procedurally generated bomb puzzles against a live countdown. One agent can see and manipulate the bomb but does not have the defusal instructions; the other has the instructions but cannot see or manipulate the bomb. Neither agent can succeed alone: success requires effective and efficient communication. Unlike turn-based proxies, GPTNT requires agents to act asynchronously and communicate in real time. GPTNT is designed to separate collaboration from reliance on memorized solutions: the instruction manual, the partner, or both can be withheld to isolate what a model derives in the moment from what it already knows. We show that GPTNT poses a substantial challenge for state-of-the-art systems: none of the closed- or open-source models we test defuses a single bomb in real time, a bar that human players clear. Through controlled experiments, we identify critical weaknesses in state tracking, efficient action under time pressure, ambiguity handling, and error recovery. We release GPTNT as a benchmark for collaborative performance that current evaluations leave unmeasured. Because it runs on the real game, GPTNT benefits from procedural generation and inherits a living modding community, allowing the benchmark to evolve as models improve rather than being solved once and retired. --- ### HDDPM: Heteroscedastic Denoising Diffusion Probabilistic Model for Quantitative Low-Count Brain PET Recovery Authors: Raymond Confidence, Udunna C. Anazodo Published: 2026-06-26 Categories: eess.IV, cs.AI arXiv: https://arxiv.org/abs/2606.28513v1 Positron emission tomography (PET) seeks to balance diagnostic quality with ra-diation dose. Low-count PET noise is non-Gaussian, non-stationary, and spatial-ly dependent. It scales directly with local activity and is shaped by iterative recon-struction and physical corrections. Standard denoising diffusion probabilistic models (DDPMs) ignore these PET properties. Their forward process adds iso-tropic, homoscedastic Gaussian noise to the target. Such an approach fails to cap-ture the realistic physical degradation generated by the imaging system. To ad-dress the above limitations, this study introduces a heteroscedastic residual diffu-sion model (HDDPM) for low-count brain PET recovery in which the forward corruption is itself intensity-aware. We designed a fixed, Poisson-based variance module to generate voxel-wise noise maps. These maps naturally place stronger noise perturbation on low-activity regions than high-activity ones, meanwhile the network predicts the low-to-standard-count residual under explicit dose-fraction conditioning. We evaluated our proposed model (HDDPM) alongside generative frameworks across three different scanners, using both internal and external da-tasets at various simulated dose levels (1% to 50%). HDDPM and isotropic DDPM showed comparable overall image quality, but HDDPM stood out in the lowest-dose (1%) external scans. It is highly reliable and significantly reduces measurement errors in both high- and low-activity regions, compared to the standard model. These results support that heteroscedastic noising with the pro-posed HDDPM is feasible, and it provides a physically motivated inductive bias for quantitative low-count PET recovery by reflecting the activity-dependent noise structure of PET. --- ### Generative AI Literacy Training Improves Intelligence Analysts' Discrimination of Real and AI-Generated Images Authors: Negar Kamali, Candice Rockell Gerstner, Jessica Hullman, Matthew Groh Published: 2026-06-26 Categories: cs.HC, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2606.28510v1 Across social and online platforms, people are increasingly exposed to AI-generated images. As a consequence, the task of distinguishing AI-generated from authentic images is becoming a central challenge for information ecosystems. While humans perform better than chance, accuracy falls short of many operational needs. Initial evidence shows that visually oriented training can improve deepfake detection but does not improve participants' ability to identify real images as real. Here, we investigate the efficacy of a brief training intervention for intelligence analysts employed by the United States government in 2024. We conducted a counterbalanced within-subject randomized experiment in which we showed participants real and AI-generated images varying in pose complexity and scene context and asked them whether each image was real or AI-generated, both before and after an expert delivered a 30-minute training that pointed out patterns in seven real and 50 AI-generated images. We collected 2,544 image-level judgments from 32 intelligence analysts. We find training increased overall accuracy by 9 percentage points (95% CI: (2.7, 15.4)) from a baseline of 72%. We find the improvement is driven by a 14.2 percentage point increase in accuracy for real images (95% CI: (0.7, 27.7)). Through a careful experimental setup that curated matched pairs of real and AI-generated images across pose complexity categories, we reveal how these trainings influence people with different levels of digital forensics and generative AI experience and identify the kind of image-based content where this training intervention appears to be most effective. Ultimately, these results provide causal evidence that a brief, structured training can improve human judgment across a diverse array of real and AI-generated images, informing organizational responses to AI-generated visual misinformation. --- ### Spectral phase transitions and trainability in neural network learning dynamics Authors: Chanju Park, Dario Bocchi, Francesco D'Amico, Biagio Lucini, Gert Aarts Published: 2026-06-26 Categories: cond-mat.dis-nn, cs.LG, hep-lat arXiv: https://arxiv.org/abs/2606.28486v1 The emergence of low-dimensional structures in the spectra of neural network weight matrices is a common empirical feature of trained models, but the dynamical origin of this phenomenon during learning remains an open problem. We formulate neural network training as the stochastic evolution of an initially random matrix ensemble, driven by stochastic gradient descent (SGD) updates that reshape the spectral bulk while amplifying signal strength. This induces a Baik-Ben Arous-Péché (BBP) transition during training, where isolated eigenvalues detach from the random bulk distribution, providing a dynamical framework for representation formation in high-dimensional learning dynamics. We demonstrate this in a solvable linear teacher-student model, where spectral evolution is analytically tractable and a phase diagram of trainability governed by the step size (or learning rate) and initial weight variance is obtained, and subsequently extend our formalism beyond the linear regime to nonlinear and stochastic settings. Numerical simulations in realistic settings support this picture, showing robust emergence of spectral alignment during training. Our results suggest that spectral analysis may provide a unified perspective of stochastic learning dynamics, linking trainability, optimisation hyperparameters, spectral phase transitions, and representation learning in neural networks. --- ### TUA-Bench: A Benchmark for General-Purpose Terminal-Use Agents Authors: Shoufa Chen, Luyuan Wang, Xuan Yang, Zhiheng Liu, Yuren Cong, Yuanfeng Ji, Feiyan Zhou, Xiaohui Zhang, Fanny Yang, Belinda Zeng Published: 2026-06-26 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2606.28480v1 As large language models and harness frameworks continue to advance, agents operating in terminals are increasingly capable of performing a broader range of general computer-use tasks beyond coding. However, existing benchmarks do not adequately evaluate general-purpose terminal computer-use agents (TUAs): general computer-use benchmarks primarily target graphical user interfaces (GUIs), whereas terminal-based benchmarks largely emphasize technical and programming-centric workflows historically native to the shell. We introduce TUA-Bench, a general-purpose benchmark for terminal-use agents. TUA-Bench includes 120 real-world tasks across five task families, covering routine digital activities-including document editing, email management, and live-web information seeking-as well as scientific and engineering workflows co-designed with PhD-level domain experts that require specialized software. This breadth distinguishes TUA-Bench from prior shell-focused or domain-specific benchmarks. Each task is manually designed, runs in a real terminal with a deterministic setup script, and is evaluated by an execution-based scoring protocol. We find that the strongest frontier agent, Claude Code with Claude Opus 4.8 max reasoning effort, achieves 65.8% overall performance, with substantial gaps across both tracks. By providing a broad and realistic evaluation of terminal-use capabilities, TUA-Bench aims to accelerate the transition from narrow, task-specific assistants to general-purpose agents capable of operating reliably across diverse digital environments. --- ### Decomposing Memorization Reduction in Privacy-Preserving Fine-Tuning of SLMs for CSIRTs Authors: Cristhian Kapelinski, Diego Kreutz Published: 2026-06-26 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2606.28479v1 CSIRTs increasingly fine tune language models on vulnerability scan records, but these records expose internal network topology and create privacy risks under regulations such as GDPR and LGPD. We present the first empirical study of how DP SGD and HMAC pseudonymization interact when fine tuning small language models with 1B to 3B parameters on structured CSIRT data. We evaluate 96 LoRA adapters across four SLMs and four training regimes, including raw fine tuning, QLoRA with large batch training, and DP SGD with epsilon equal to 2 and 8. We also audit memorization using 20 planted canaries, four extraction attacks, and a dual attack targeting HMAC pseudonymized identifiers. Our results show three main findings. First, matched update controls reproduce the observed reduction in memorization by reducing the number of optimizer updates alone, accounting for 66 percent to 132 percent of the measured effect, with a mean of 100 percent across three seeds and four models. In this setting, DP SGD provides the formal privacy guarantee but does not produce additional measurable reductions in memorization. Second, HMAC pseudonymization removes the original identifiers from the exposure surface, reducing exposure by 40 percent to 61 percent, while pseudonymized identifiers remain close to the expected random baseline and do not become a secondary memorization target. Third, F1 scores remain between 0.19 and 0.28 across all 96 adapters using four shot prompting, indicating that, under the evaluated training budget, 1B to 3B SLMs do not achieve operationally useful performance. --- ### Improvement of Robot's Simultaneous Localization and Mapping Using an Effective Transformation to Achieve Linear Model Authors: Seyed Farzad Bahreinian, Maziar Palhang, Mohammad Reza Taban, Hasan Enami Eraghi Published: 2026-06-26 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2606.28475v1 Nowadays mobile robots have wide engineering applications. Simultaneous localization and mapping (SLAM) is an important task of these robots. The major and common algorithms used for this task are based on extended Kalman filter (EKF). One of the main problems in EKF-based SLAM is its divergence. The nonlinearity of motion and observation models and linearization error are the main reasons for the divergence. There have been some efforts to address this problem with limited success. In this paper, by applying a simple compass and using an effective transformation, we transform the non-linear state space model into a linear model. Then, by applying the original KF to this model, we reach a new method, which is called LMKF SLAM. We show that the LMKF SLAM is significantly superior to the state-of-the-art methods, especially EKF-based SLAMs, both in accuracy, convergence, and computational complexity. The proposed method is also more stable with respect to the uncertainty of sensors values and changes in system parameters. Experimental results verify these points. --- ### Anatomy-Grounded Synthetic Coronary Angiography for Geometry-Informed Multi-View Matching Authors: In Kyu Lee, Sumin Seo, Jaesik Min Published: 2026-06-26 Categories: eess.IV, cs.CV arXiv: https://arxiv.org/abs/2606.28474v1 Accurate correspondence matching across multiple angiographic views is the prerequisite for 3D coronary reconstruction and interventional guidance. However, the development of robust deep learning models for this task has been stifled by a fundamental data bottleneck. Obtaining ground truth for matching tasks in angiography pairs is prohibitively expensive and hard to scale. To overcome this barrier, we introduce a physically-grounded data generation framework that synthesizes high-fidelity Digital Reconstructed Radiographs (DRRs) from 3D Coronary CT Angiography (CCTA) volumes. Our framework generates dense, highly accurate 3D-to-2D projection labels by simulating realistic C-arm acquisition geometry on patient anatomy at zero human cost. Leveraging this dense supervision, we propose a Geometry-Informed Matching Module (GIMM) that integrates global feature and anatomical structure into correspondence learning. Unlike real angiography where assessment relies on subjective human annotation, our dataset provides 2D correspondence labels with paired images, allowing human-free evaluation. We comprehensively evaluate our method on the proposed CT-derived DRR dataset and demonstrate improvements over other matching baseline models. --- ### Data and Evaluation Closed-Loop for Model Capability Enhancement Authors: Zhixuan Li, Jiangan Yuan, Han Xu Published: 2026-06-26 Categories: cs.AI arXiv: https://arxiv.org/abs/2606.28471v1 Model capability is the central variable in LLM pre-training, yet is never observed directly: data shapes it prospectively, while evaluation reveals it only retrospectively, compressing samples, prompts, decoding, and scoring rules into one noisy score. Practical optimization runs this backward: a failure is observed first, and the engineer must infer the corpus fix. The two sides speak incompatible vocabularies -- benchmark names and per-sample correctness versus data sources, domains, and quality labels -- so this inference is usually intuition, not method. We close this gap with the \emph{capability slice}: a group of evaluation samples sharing background condition, task type, solving operation, and output constraint -- precise enough to localize a single weakness yet stable enough to survive aggregation, unlike a benchmark name, too coarse, or a single sample, too noisy. Built around this unit, an evaluation taxonomy, a non-instruction data taxonomy, and mapping rules form a closed loop turning a benchmark-level failure into a targeted, testable data intervention. We test this loop on two case studies pulling in opposite directions. First, the loop rules the data out: continued pre-training drives BBH down by $-46.82\%$, but diagnosis traces this to a single masked \texttt{\textless EOS\textgreater} loss rather than weakened reasoning; restoring it recovers BBH to $66.44$, above the original checkpoint, without changing the data. Second, the loop rules the data in: a persistent math-reasoning weakness is decomposed by solving operation into specific failing combinations, and a weakness-targeted sampling procedure built from it lifts AIME2025/AIME2026 Pass@128 from $6.67$/$0.00$ to $26.67$ each. The same unmodified loop reaches opposite, correct verdicts in both cases, showing the evaluation-to-data inference can be routine, auditable, and experimentally validated rather than intuitive. --- ### Modelling Emotional Memory in Children with Tensor Networks Authors: Henry Groves, Lucia F. Jackson, Barbara-Anne Robertson, Jonte R. Hance Published: 2026-06-26 Categories: cs.LG, q-bio.NC, quant-ph arXiv: https://arxiv.org/abs/2606.28470v1 We demonstrate how emotional valence influences the order-dependent structure of children's recognition memory: correct recall of a sequence of emotionally-valenced toys depended not just on the valence of a given toy itself, but also on the valence of the toys shown before and after it. Whilst standard psychological models confirm that order-dependence differs across an event (a set of toys shown in sequence), accuracy is low and the model does not reflect how memory for an emotional object influences others in the set. A classical tensor network model factoring in valence is able to achieve a 77.98\% accuracy in modelling the results of the study. While not strictly a ``quantum cognition'' model, this massive increase in accuracy shows the value of quantum-inspired methods for modelling order-dependent phenomena, such as emotional memory. Further, the task protocol we introduce presents a novel, real-world tool for exploring emotional temporal memory in children for analysis using classical and quantum-like models of cognition. --- ### An Agentic AI Pipeline for Appliance-Level Energy Anomaly Detection and LLM-Driven Recommendations Authors: Dihia Falouz, Aida Douaibia, Amine Bechar, Youssef Elmir, Abbes Amira, Adel Oulefki Published: 2026-06-26 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2606.28467v1 Appliance-level energy monitoring in office buildings produces noisy alerts that non-expert facility managers struggle to use. This paper proposes an end-to-end agentic pipeline that combines deep time-series forecasting, variational anomaly detection, and LLM-based reasoning to generate prioritized, actionable maintenance recommendations. The system tracks seven office appliances using a hybrid Singular Spectrum Analysis (SSA) and Long Short-Term Memory (LSTM) forecasting model, and applies a per-appliance LSTM Variational Autoencoder (VAE) with attention to flag abnormal daily consumption episodes. A three-stage LangChain pipeline begins with a Context Agent that always retrieves three core RAG sources (model reliability, hourly baseline, and expert knowledge) and conditionally adds up to three more (forecast context, anomaly history, global baseline) based on event characteristics, capped at eight reasoning steps. A Diagnosis Agent converts the evidence into a structured JSON diagnosis, and a Report Agent renders a human-readable narrative. A reflective memory layer incorporates operator feedback. The dashboard shows real-time 30-minute forecasts, intraday consumption, the previous day anomaly report, and a feedback form. We evaluate the forecasting model, anomaly detector with appliance-specific thresholds, and LLM reasoning on a 16-scenario benchmark including sustained and transient spikes, unexpected shutdowns, and systemic events, comparing five LLM backends under static vs. dynamic retrieval. Dynamic retrieval matches full static retrieval across all backends while cutting average context from six to three-six sources per event. The best backend scores 90.4/100 with a 100% pass rate at a 70-point threshold, and a fully local 7B-parameter model passes all 16 scenarios. --- ### SVC-Probe: A Framework for Evaluating Perturbation Generalization in Spatial Foundation-Model Embeddings Authors: Jake Y. Chen, Huu Phong Nguyen, Fuad Al Abir, Ehsan Saghapour Published: 2026-06-26 Categories: q-bio.QM, cs.AI arXiv: https://arxiv.org/abs/2606.28465v1 This work examines perturbation generalization in spatial foundation-model embeddings derived from fluorescence microscopy images. Although these models can discriminate drug conditions accurately, it remains unclear whether the learned representations reflect patterns consistent with expected perturbation axes that transfer across drugs. We introduce SVC-Probe, a perturbation-aware framework that combines Subcellular Embedding Atlas Stability, Mondrian Neighborhood Graphs, and a Foundation Model Perturbation Probe to assess embedding stability, neighborhood rewiring, and centroid prediction under drug treatment. Applied to the CM4AI MDA-MB-468 chemical-perturbation atlas comprising 462 antibody labels and SubCell 1536-dimensional embeddings, SVC-Probe demonstrates that 98.6% three-way condition accuracy does not correlate with reliable cross-drug prediction, with cosine similarity diminishing from 0.944 in-domain to 0.30 under leave-one-drug-out evaluation, constituting a two-drug stress test rather than a general benchmark. Null calibration indicates that raw residual-turnover coupling is largely influenced by generic embedding structure, whereas a drug-specific signal emerges under vorinostat and is consistent with chromatin-related reorganization. In contrast, the paclitaxel axis is not robustly reconstructed, likely due to sparse coverage of microtubule-associated proteins. Together, these results introduce and demonstrate a reusable diagnostic framework for stress-testing spatial virtual-cell representations and indicate that perturbation generalization may serve as a stricter and more informative benchmark than baseline condition discrimination. --- ### Singular Learning and Occam's Razor in Deep Monomial Networks Authors: Kathlén Kohn, Giovanni Luca Marchetti, Farhan Shabir, Vahid Shahverdi, Weisheng Wang Published: 2026-06-26 Categories: cs.LG arXiv: https://arxiv.org/abs/2606.28464v1 In the optimization of neural networks, gradient dynamics are influenced by critical points that arise from the model's architecture. These critical points occur where the Jacobian of the model's parametrization is rank-deficient, and are the most pronounced singularities studied in Singular Learning Theory. We investigate such points in deep fully-connected networks with monomial activations via tools from polynomial algebra such as Mason's Theorem. We show that, for sufficiently large activation degree, criticality occurs precisely at subnetworks, i.e., at parameter configurations where some neurons are inactive or redundant. This offers a mathematical perspective on the implicit bias in deep neural networks, explaining the tendency of these models to converge toward simpler functions. --- ### Counterfactual Residual Data Augmentation for Regression Authors: Hossein Mohebbi, Oliver Schulte, Ke Li, Pascal Poupart Published: 2026-06-26 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.28460v1 Data-driven modeling in real-world regression tasks often suffers from limited training samples, high collection costs, and noisy observations. Inspired by the impact of data augmentation in vision and language, we propose a novel Counterfactual Residual Data Augmentation (CRDA) technique for tabular regression. Our key insight is that once a regressor has modeled the systematic component of the data, the remaining noise can be viewed as an invariant residual that remains stable under small perturbations of carefully selected features. We exploit this residual invariance to generate new, yet realistic, training samples, effectively expanding the dataset without requiring additional real data. Our method is model-agnostic and readily applicable to various types of regressors. In experiments across datasets from a variety of benchmark repositories, on average, CRDA reduces an MLP Regressor's MSE by 22.9% and an XGBoost Regressor's MSE by 6.4%. When compared to existing state-of-the-art data generators and augmentation techniques, CRDA consistently outperforms in MSE reduction. By adding principled counterfactual variations to the training data, our method offers a simple and efficient remedy for noise-prone, small-sample regression settings. --- ### scKDGM: KAN-guided Dynamic Graph Masked Learning for Single-Cell RNA-seq Clustering Authors: Jun Tang, Pengwei Hu, Sicong Gao, Jie Guo, Lun Hu, Xin Luo Published: 2026-06-26 Categories: cs.LG, q-bio.GN arXiv: https://arxiv.org/abs/2606.28459v1 Single-cell RNA sequencing (scRNA-seq) clustering is essential for identifying cell types, but high dimensionality, sparsity, dropout, and technical noise hinder robust expression representation and cell graph construction. Existing masked autoencoders mainly use expression recovery for feature reconstruction, while graph clustering methods usually depend on fixed KNN graphs and do not feed recovered expression back into graph optimization. We propose scKDGM, a KAN-guided dynamic graph masked learning framework for scRNA-seq clustering. scKDGM uses graph-aware distribution preserving gene masking (GDP-Mask) to perturb cell identity, a KAN-based TAKGCN encoder to learn masked-view representations, mask-guided expression recovery to construct a dynamic graph, and cross-view contrastive learning to transfer recovery signals into topology updates. A ZINB loss models overdispersion and zero inflation. Experiments on 12 real scRNA-seq datasets show that scKDGM outperforms 10 baselines in average NMI and ARI. ---