# Research Papers — 2026-05 · bundle page 267 of 326 > Full-text markdown of papers #13301–13350 of 16266 published in 2026-05, newest first. AI-readable mirrors of the full 16266-paper 2026-05 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-05/page/267 [← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/266.md) [Next bundle →](https://aipod.dev/papers/archive/2026-05/page/268.md) ### Rollback-Free Stable Brick Structures Generation Authors: Chenhui Xu, Ziyue Bai, Fuxun Yu, Heng Huang, Jinjun Xiong Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06947v1 While autoregressive models have advanced 3D generation, creating physically stable brick structures remains a challenge due to the strict requirements of gravity and interconnectivity. Existing approaches rely on external physical simulators during inference to perform rejection sampling and brick-by-brick rollbacks, which severely bottlenecks efficiency. To address this, we propose a reinforcement learning paradigm that shifts physical validity enforcement from test-time correction to training-time policy optimization. By utilizing assembly-level rewards, the model optimizes for collision avoidance, global connectivity, structural interlocking, and shape conformity. This paradigm allows the model to internalize physical priors, enabling the first rollback-free generation of stable brick structures. Experimental results demonstrate that our approach achieves state-of-the-art generation quality while accelerating inference speed by orders of magnitude. Our code and dataset are available at https://github.com/miniHuiHui/STABLE. Our models are available at https://huggingface.co/miniHui/STABLE. --- ### Adaptive Memory Decay for Log-Linear Attention Authors: Yaxita Amin, Helen Zichen Li, Mengfan Zhang, Samet Ayhan Published: 2026-05-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.06946v1 Sequence models face a fundamental tradeoff between memory capacity and computational efficiency. Transformers achieve expressive context modeling at quadratic cost, while linear attention and state-space models run in linear time by compressing context into a fixed-size hidden state, inherently limiting recall. Log-linear attention navigates this tradeoff by organizing memory across a Fenwick tree hierarchy, growing its hidden state logarithmically with sequence length at log-linear compute cost. However, its memory decay parameter λ is fixed and independent of the input, assigning uniform weights across all hierarchy levels regardless of the content, which introduces unnecessary rigidity. We propose learning λ directly from the input via a lightweight two-layer MLP, producing per-token, per-level decay that adapts to content rather than position. A softplus activation lets each Fenwick tree level scale independently, avoiding the inter-level competition that softmax introduces. This modification preserves log-linear complexity exactly and adds negligible parameter overhead. We evaluate on associative recall, selective copying, and language modeling, finding that input-dependent decay consistently outperforms the baseline, with the largest gains in long-range memory settings where baseline λ degrades or collapses entirely. --- ### ProtoSSL: Interpretable Prototype Learning from Unlabeled Time-Series Data Authors: Steven Song, Sahil Sethi, Brett Beaulieu-Jones, Robert L. Grossman Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06943v1 In time-series domains where both predictive performance and interpretability are essential, deep neural networks achieve strong results but provide limited insight into how their predictions are made. Projection-based prototype networks address this limitation by grounding predictions in similarity to representative training examples, enabling case-based explanations and global prototype inspection. However, existing approaches rely on label supervision, tying prototypes to a specific task and requiring large labeled datasets. We introduce ProtoSSL, a novel framework for learning interpretable, projection-based prototypes from unlabeled time-series data and adapting them to downstream tasks. Our key idea is to separate motif discovery from label alignment. ProtoSSL first learns a reusable prototype bank using a self-supervised objective applied directly to prototype activations, and then aligns these prototypes to downstream tasks through an efficient assignment procedure. Across six electrocardiography (ECG) datasets, ProtoSSL improves label efficiency, outperforming supervised prototype baselines in low-data regimes with as few as 256 labeled examples; with fine-tuning, ProtoSSL outperforms supervised prototype baselines at full dataset scale. In a human evaluation study, ProtoSSL produces prototypes and prototype-based explanations that are judged more favorably than those learned with direct label supervision. We further show that the framework extends to audio classification. Thus, ProtoSSL enables both learning generalizable prototypes from unlabeled data before the downstream label space is known, and subsequent assignment of interpretable, projection-grounded prototypes to new time-series tasks. --- ### Causal-Aware Foundation-Model for Bilevel Optimization in Discrete Choice Settings Authors: Shivaram Subramanian, Zhengliang Xue, Markus Ettl, Yingdong Lu, Jayant Kalagnanam Published: 2026-05-07 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2605.06941v1 We introduce a causal aware foundation-model framework for real time optimal decision making in discrete choice environments. We propose a constrained triple-head price optimization (C3PO) network to solve a bilevel decision problem in which a service provider selects an optimal assortment while heterogeneous users make personalized acceptance or rejection choices optimizing their own personalized preferences. C3PO integrates imitation learning of prices, multi-task learning of revenue responses, and in context learning of price elasticity to generate pricing recommendations while adhering to business constraints. During inference, frontier model prompting retrieves an enhanced elasticity prior for new products from behavioral economics literature, improving pricing effectiveness. We demonstrate strong in context learning performance using simulated, synthetic, and real-world datasets. C3PO is trained on simulated data generated from multiple classical discrete choice models in economics. The model is trained on data comprising simulated customer segments and counterfactual action and outcome pairs and evaluated on randomly generated choice environments with no access to the underlying preference structure. The trained model consistently improves the pricing KPIs, with gains increasing as customer price sensitivity increases. We also deploy the tuned foundation model for optimal pricing in real-world applications such as healthcare, tender pricing, airline ancillary pricing, and other domains, achieving substantial gains across multiple products, markets, and divisions. --- ### MultiSoc-4D: A Benchmark for Diagnosing Instruction-Induced Label Collapse in Closed-Set LLM Annotation of Bengali Social Media Authors: Souvik Pramanik, S. M. Riaz Rahman Antu, Shak Mohammad Abyad, Md. Ibrahim Khalil, Md. Shahriar Hussain Published: 2026-05-07 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.06940v1 Annotation automation via Large Language Models (LLMs) is the core approach for scaling NLP datasets; however, LLM behavior with respect to closed-set instructions in low-resource languages has not been well studied. We present MultiSoc-4D, a Bengali social media dataset benchmark, which contains 58K+ social media comments from six sources annotated along four dimensions: category, sentiment, hate speech, and sarcasm. By employing a structured pipeline where ChatGPT, Gemini, Claude, and Grok individually annotate separate partitions, while sharing a common validation set of 20%, we diagnose LLM behavior systematically. We discover a prevalent phenomenon called "instruction-induced label collapse", wherein LLMs show a systematic preference towards fallback labels (Other, Neutral, No), leading to high agreement rates but under-detection of minority categories. For example, we find that LLMs failed to detect 79% and 75% of instances with hateful and sarcastic content compared to a human-calibrated reference. Furthermore, we prove that it represents a "label agreement illusion", statistically validated via almost null Fleiss' Kappa ($κ\approx -0.001$) on sarcasm detection. Across 40+ LLMs, we benchmark this annotation bias propagation within the training pipeline, regardless of architectural differences. We release MultiSoc-4D as a diagnostic benchmark for annotation biases in Bengali NLP. --- ### Bias and Uncertainty in LLM-as-a-Judge Estimation Authors: James Fiedler Published: 2026-05-07 Categories: cs.LG, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2605.06939v1 LLM-as-a-Judge evaluation has become a standard tool for assessing base model performance. However, characterizing performance via the naive estimator, i.e., raw judge outputs, is systematically biased. Recent work has proposed estimators to correct this bias, but their reliability depends critically on judge quality and, for model comparisons, on calibration stability. Sharing calibration across compared models is practically attractive but can introduce severe bias, including cases where the comparison estimate points in the wrong direction with high apparent confidence. We study these failure modes through analytical results, simulations over judge quality ($J$) and cross-model calibration instability ($ΔJ$), and a real-data MMLU-Pro case study with sign reversal. We propose $J$ and $ΔJ$ as diagnostics for when corrected estimates, especially shared-calibration comparisons, are likely unreliable, and provide reporting guidance for LaaJ evaluation. --- ### A Nonlinear Singular Value Theory for Neural Networks Authors: Brian Charles Brown, Mauricio Munoz, Robert Bridges, David Grimsman, Sean Warnick Published: 2026-05-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.06938v2 Recently Brown et al. (2025) established a singular value decomposition (SVD) for maps (especially nonlinear) satisfying certain norm conditions. We prove that most modern neural architectures admit this nonlinear SVD (NLSVD) representation---with no change in input--output behavior---and enumerate the classes covered. In this factorization the network is a left-invertible nonlinear map followed by a final linear layer. Moreover, the left-invertible factor is norm-preserving, so distances in the embedding (activations before the final linear layer) calibrate directly to distances in input space. We introduce a flexible architecture that yields an explicit decomposition at training time, a data-driven algorithm for estimating the representation from trained models, and the mathematical foundations for nonlinear analogues of row and null spaces in neural networks. Empirical case studies illustrate uses of the theory for latent-space pullback (visualization and data generation), bias detection, and membership-inference robustness under training. Altogether, these foundations support new approaches to core problems in neural-network analysis. --- ### A Reproducible Optimisation Protocol for Calibrating Prompt-Based Large Language Model Workflows in Evidence Synthesis Authors: Teo Susnjak Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06937v1 This methods article presents a reproducible calibration workflow for prompt-based large language models (LLMs) in structured evidence-synthesis tasks. The method separates the rules that define the scientific task from the mutable prompt harness that frames and applies them. It optimises that harness against labelled or reference examples and an explicit task metric, then preserves the calibrated workflow as an inspectable artefact with its specification, metric, settings, and evaluation traces. The example code instantiates the protocol with DSPy and GEPA tools, but the underlying logic can transfer to other prompt-optimisation frameworks that support structured task definitions, metric-guided search, and artefact reuse. Title and abstract screening is the worked validation case because it provides labelled benchmark data and clear evaluation metrics. The demonstrated workflow uses a smaller student LLM for performing the scientific task execution and a larger reflection LLM to steer the prompt optimisation process during calibration. This work shows compilation, artefact round-tripping, and how optimisation budget affects a smaller student model. --- ### Bridging the Last Mile of Circuit Design: PostEDA-Bench, a Hierarchical Benchmark for PPA Convergence and DRC Fixing Authors: Pengju Liu, Nuo Xu, Jinwei Tang, Yu Cao, Caiwen Ding Published: 2026-05-07 Categories: cs.AR, cs.AI, cs.MA arXiv: https://arxiv.org/abs/2605.06936v3 LLM-based agents are increasingly applied to the "last mile" of Electronic Design Automation (EDA): repairing residual sign-off Design Rule Check (DRC) violations and converging Power-Performance-Area (PPA) targets after tool runs. Existing EDA-LLM benchmarks, however, omit DRC fixing entirely and rely on flat hierarchies tied to a single toolchain. We introduce PostEDA-Bench, a hierarchical benchmark with 145 tasks across DRC-Essential, DRC-Reasoning, PPA-Mono, and PPA-Multi, supported by EDA toolchains with machine-checkable evaluation. Across eight commercial and open-source LLMs under multiple agent scaffolds, we find that agents handle synthetic DRC-Essential and single-objective PPA-Mono reasonably well but degrade sharply on the more practical DRC-Reasoning, where the best success rate is 36.66%, and PPA-Multi, where the best success rate is 20.00%; vision augmentation consistently enhances DRC-Bench; and trade-off reasoning, rather than knob knowledge, is the dominant PPA-Multi bottleneck. --- ### Learned Lyapunov Shielding for Adaptive Control Authors: Giansalvo Cirrincione, Adriano Fagiolini Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06934v1 We augment the Slotine--Li adaptive controller for Euler--Lagrange systems with three learned components: a structured-quadratic Lyapunov function \(V_ψ\) whose positive-definiteness follows from a Cholesky parameterization, a residual Soft Actor--Critic policy that adds bounded torque corrections to the analytic baseline, and a physics-informed neural network that estimates unmodeled dynamics. A closed-form safety filter, derived from the single affine constraint \(\dot V_ψ+ αV_ψ\le 0\), projects every policy output onto the safe set without requiring an online QP solver. We prove: global feasibility of the filter under a drift-decay condition on the control-degeneracy set; exponential stability under exact shielding, with a robust extension whose margin depends on the PINN approximation error; almost-sure convergence of the three-timescale policy--certificate--multiplier updates to a KKT point; and a PAC generalization bound for the certificate over compacts. On a 2-DOF manipulator with nonlinear friction and variable payload, the learned certificate accounts for most of the empirical gain: tracking error drops by 41\% on nominal friction and 24\% on aggressive friction at the centroid of the training distribution. A 7-DOF scalability study on a Franka Emika Panda confirms clean convergence of the full pipeline at industrial scale, identifies the conditions under which gains over exact model-based baselines should and should not be expected, and documents a warm-start pathology of the learned certificate that has practical implications for deployment. --- ### MAGIQ: A Post-Quantum Multi-Agentic AI Governance System with Provable Security Authors: Sepideh Avizheh, Tushin Mallick, Alina Oprea, Cristina Nita-Rotaru, Reihaneh Safavi-Naini Published: 2026-05-07 Categories: cs.LG, cs.CR, cs.MA arXiv: https://arxiv.org/abs/2605.06933v2 Our computing ecosystem is being transformed by two emerging paradigms: the increased deployment of agentic AI systems and advancements in quantum computing. With respect to agentic AI systems, one of the most critical problems is creating secure governing architectures that ensure agents follow their owners' communication and interaction policies and can be held accountable for the messages they exchange with other agents. With respect to quantum computing, existing systems must be retrofitted and new cryptographic mechanisms must be designed to ensure long-term security and quantum resistance. In fact, NIST recommends that standard public-key cryptographic algorithms, including RSA, Diffie-Hellman (DH), and elliptic-curve constructions (ECC), be deprecated starting in 2030 and disallowed after 2035. In this paper, we present MAGIQ, a framework for policy definition and enforcement in multi-agent AI systems using novel, highly efficient, quantum-resistant cryptographic protocols with proven security guarantees. MAGIQ (i) allows users to define rich communication and access-control policy budgets for agent-to-agent sessions and tasks, including global budgets for one-to-many agent sessions; (ii) enforces such policies using post-quantum cryptographic primitives; (iii) supports session-based enforcement of policies for agent-to-agent and one-to-many agent sessions; and (iv) provides accountability of agents to their users through message attribution. We formally model and prove the correctness and security of the system using the Universal Composability (UC) framework. We evaluate the computation and communication overhead of our framework and compare it with the state-of-the-art agentic AI framework SAGA. MAGIQ is a first step toward post-quantum-secure solutions for agentic AI systems. --- ### Target-Aware Data Augmentation for SAT Prediction Authors: Eshed Gal, Uri Ascher, Eldad Haber Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06931v1 Learning-based approaches to NP-hard problems have shown increasing promise, but their progress is fundamentally constrained by the high cost of generating labeled training data. In domains such as Boolean satisfiability (SAT), standard pipelines rely on solver-in-the-loop labeling, which scales poorly with problem size and limits the amount of usable supervision. This bottleneck hinders the broader goal of leveraging machine learning to capture structure in hard combinatorial problems. In this work, we propose a target-aware, solver-free data generation framework for SAT that produces correctly labeled SAT and UNSAT instances by construction, eliminating the need for expensive solver calls. Our method aligns generated instances with the structural properties of a target benchmark, making synthetic data effective for downstream learning. We further develop a linear-programming-aware graph neural network (LPGNN) architecture that incorporates constraint-violation residuals into message passing, enabling the model to exploit underlying optimization structure. Together, these contributions support a data-centric paradigm for learning on NP-hard problems, where scalable, task-aligned data generation is as critical as model design. Our approach yields orders-of-magnitude speedups in data generation, demonstrating that benchmark-aligned synthetic data can effectively augment solver-labeled datasets for GNN-based SAT prediction. --- ### Physics-Based Flow Matching for Full-Field Prediction of Silicon Photonic Devices Authors: Joseph Quaratiello, Anthony Rizzo Published: 2026-05-07 Categories: physics.optics, cs.LG arXiv: https://arxiv.org/abs/2605.06929v1 Designing photonic integrated circuits requires accurate electromagnetic field simulations, which remain computationally expensive even for simple device geometries. We present PIC-Flow, a generative neural surrogate that predicts electromagnetic field distributions for photonic devices given their geometry and operating wavelength as an alternative to costly finite-difference time-domain (FDTD) simulations. Our approach combines three key ideas: (i) conditional flow matching as the generative framework, learning a velocity field that transports Gaussian noise to physically valid field solutions; (ii) a real-valued U-Net operating on split real and imaginary field channels; and (iii) physics-constrained training through a Helmholtz residual loss enforcing $\nabla^2 E_z + k_0^2 \varepsilon E_z = 0$. We introduce an interface-aware masking scheme for the Helmholtz residual that excludes dielectric boundary pixels where finite-difference stencil errors dominate, yielding a physically meaningful compliance metric. The data set consists of 22,500 ground-truth FDTD simulations split evenly between multimode interferometers, Y-branches, and directional couplers at $λ=1.55\,μ$m in an 80/10/10 split between training, validation, and test sets. We evaluate ablations on the network against the held out test devices and also show that the model generalizes to held out device classes such as S-bends, tapers, and cascaded Y-branches. Rather than a drop-in replacement for FDTD, this work establishes a foundation that, with broader data coverage, more compute, and further training optimization, could scale toward broadband, device-agnostic field prediction with dramatically improved runtime for rapid design-space exploration of complex photonic devices and circuits. --- ### XiYOLO: Energy-Aware Object Detection via Iterative Architecture Search and Scaling Authors: Tony Tran, Richie R. Suganda, Bin Hu Published: 2026-05-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.06927v1 Object detection on heterogeneous edge devices must satisfy strict energy, latency, and memory constraints while still providing reliable perception for downstream autonomy. Existing energy-aware NAS methods often target limited deployment settings, while real energy remains difficult to optimize because it is highly device-dependent and costly to measure. We address these challenges with an energy-adaptive framework that combines an energy-aware XiResOFA search space, a two-stage energy estimator, and iterative search to identify a single energy-efficient base architecture. We then apply compound scaling to transform this base design into the XiYOLO family across deployment budgets, enabling interpretable accuracy-energy tradeoffs under sparse hardware measurements. Experiments on PascalVOC, COCO, and real-device deployment show that XiYOLO achieves a stronger energy-accuracy tradeoff than YOLO baselines. On PascalVOC, the medium XiYOLO model reaches 86.15 mAP50 while reducing energy relative to YOLOv12m by 20.6% on GPU and 35.9% on NPU. On COCO, XiYOLO reduces energy relative to YOLOv12 by up to 53.7% on GPU and 51.6% on NPU at the small scale. The proposed two-stage estimator also improves sample efficiency over a joint predictor under few-shot adaptation with only 2-20 target-device samples. --- ### A$^2$RD: Agentic Autoregressive Diffusion for Long Video Consistency Authors: Do Xuan Long, Yale Song, Min-Yen Kan, Tomas Pfister, Long T. Le Published: 2026-05-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.06924v1 Synthesizing consistent and coherent long video remains a fundamental challenge. Existing methods suffer from semantic drift and narrative collapse over long horizons. We present A$^2$RD, an Agentic Auto-Regressive Diffusion architecture that decouples creative synthesis from consistency enforcement. A$^2$RD formulates long video synthesis as a closed-loop process that synthesizes and self-improves video segment-by-segment through a Retrieve--Synthesize--Refine--Update cycle. It comprises three core components: (i) Multimodal Video Memory that tracks video progression across modalities; (ii) Adaptive Segment Generation that switches among generation modes for natural progression and visual consistency; and (iii) Hierarchical Test-Time Self-Improvement that self-improves each segment at frame and video levels to prevent error propagation. We further introduce LVBench-C, a challenging benchmark with non-linear entity and environment transitions to stress-test long-horizon consistency. Across public and LVBench-C benchmarks spanning one- to ten-minute videos, A$^2$RD outperforms state-of-the-art baselines by up to 30% in consistency and 20% in narrative coherence. Human evaluations corroborate these gains while also highlighting notable improvements in motion and transition smoothness. --- ### In-Context Credit Assignment via the Core Authors: Keegan Harris, Siddharth Prasad, Asher Trockman Published: 2026-05-07 Categories: cs.GT, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.06920v2 We propose incentive-aligned mechanisms for in-context credit assignment: the task of assigning credit for AI-generated content among creators whose intellectual property appears in the context window. Our approach is based on the least core solution concept from cooperative game theory, which distributes value in a way that is as stable as possible by ensuring that no subset of creators is significantly under-compensated relative to the value they could generate on their own. We develop algorithms for approximating the least core, which leverage novel routines for constraint seeding and constraint separation. On a web retrieval credit assignment task, we find that our approaches approximate the least core using an order of magnitude fewer LLM calls compared to alternative methods. --- ### In-Context Credit Assignment via the Core Authors: Keegan Harris, Siddharth Prasad, Asher Trockman Published: 2026-05-07 Categories: cs.GT, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.06920v1 We propose incentive-aligned mechanisms for in-context credit assignment: the task of assigning credit for AI-generated content (e.g. code, news articles, short-form videos) among creators whose intellectual property appears in the context window. Our approach is based on the least core solution concept from cooperative game theory, which distributes value in a way that is as stable as possible by ensuring that no subset of creators is significantly under-compensated relative to the value they could generate on their own. We develop algorithms for approximating the least core, which leverage novel routines for constraint seeding and constraint separation. On a web retrieval credit assignment task, we find that our approaches are capable of approximating the least core using orders of magnitude fewer LLM calls compared to alternative methods. --- ### Can LLMs Take Retrieved Information with a Grain of Salt? Authors: Behzad Shayegh, Mohamed Osama Ahmed, Fred Tung, Leo Feng Published: 2026-05-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.06919v1 Large language models have demonstrated impressive retrieval-augmented capabilities. However, a crucial area remains underexplored: their ability to appropriately adapt responses to the certainty of the retrieved information. It is a limitation with real consequences in high-stakes domains like medicine and finance. We evaluate eight LLMs on their context-certainty obedience, measuring how well they adjust responses to match expressed context certainty. Our analysis reveals systematic limitations: LLMs struggle to recall prior knowledge after observing an uncertain context, misinterpret expressed certainties, and overtrust complex contexts. To address these, we propose an interaction strategy combining prior reminders, certainty recalibration, and context simplification. This approach reduces obedience errors by 25% on average, without modifying model weights, demonstrating the efficacy of interaction design in enhancing LLM reliability. Our contributions include a principled evaluation metric, empirical insights into LLMs' uncertainty handling, and a portable strategy to improve context-certainty obedience across diverse LLMs. --- ### Generalising Travel Time Prediction To Varying Route Choices In Urban Networks Authors: Łukasz Gorczyca, Kacper Drozd, Michał Bujak, Rafał Kucharski Published: 2026-05-07 Categories: cs.MA, cs.LG arXiv: https://arxiv.org/abs/2605.06918v1 Previous methods that predict system-wide travel time, predominantly grounded in graph neural networks, remain limited to typical and recurring demand patterns. While they successfully predict future congestion following daily commute, they inherently approximate a single demand realisation and fail to capture varying route choices. In this work, we propose a Generalised Travel Time Predictor (GenTTP) that successfully differentiates route choices and offers accurate flow and travel time predictions. Our framework learns to uncover complex spatiotemporal traffic patterns and microscopic relationships between route choices and the resulting travel times. This addresses a critical gap: the lack of travel time prediction models that generalise across varying route assignments, where the same demand can produce substantially different network-wide outcomes depending on how travellers are distributed over available paths. --- ### Tyche: One Step Flow for Efficient Probabilistic Weather Forecasting Authors: Fan Xu, Yuan Gao, Kun Wang, Rui Su, Fenghua Ling, Hao Wu, Wanli Ouyang Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06916v1 Probabilistic weather forecasting requires not only accurate trajectories, but calibrated distributions over plausible atmospheric futures. Recent data-driven systems have achieved remarkable deterministic skill, and diffusion-based ensemble forecasters have substantially improved sample realism and uncertainty quantification. However, their inference cost scales with forecast horizon, ensemble size, and the number of denoising steps required for each transition, making large operational ensembles expensive. To address this, we present Tyche, a one-step conditional flow model for efficient probabilistic weather forecasting. Tyche models the conditional forecast distribution with a destination-aware average-velocity flow that maps Gaussian noise directly to future weather states in a single function evaluation (1-NFE). To make this one-step transport learnable in high-dimensional geophysical fields, we derive a JVP-regularized rectification objective that enforces temporal self-consistency across source and destination flow timesteps without explicitly forming Jacobians. The transport field is parameterized by an isotropic Swin-style transformer that preserves fine-scale spatial structure while remaining scalable on global grids. To improve ensemble reliability under autoregressive forecasting, we further introduce a rollout-based finetuning stage with curriculum CRPS calibration supervision. Experiments on ERA5 at 1.5$^\circ$ and 6-hour resolution show that our Tyche, using merely a single NFE, matches or exceeds the forecast skill and calibration of state-of-the-art multi-step generative baselines and the operational ECMWF IFS ensemble. --- ### LLMs are not (consistently) Bayesian: Quantifying internal (in)consistencies of LLMs' probabilistic beliefs Authors: Chacha Chen, Matthew Jörke, Adam Goliński, Masha Fedzechkina, Guillermo Sapiro, Sinead Williamson, Nicholas Foti Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06915v2 Modern AI systems are being deployed in complex domains such as medicine, science, and law, where it is important that they not only produce correct answers, but also represent and update uncertain beliefs about the world as new evidence arrives. We introduce the novel technique of studying LLMs as information processing rules and utilize the information processing gap to study the internal (in)consistencies of how LLMs update their probabilistic beliefs from evidence. Our extensive experiments evaluate multiple approaches in which LLMs can incorporate evidence into their beliefs. Some of these approaches produce (nearly) Bayesian updates; others seem to use a learned heuristic. Surprisingly, the non-Bayesian heuristic updates often outperform exact Bayesian computation in terms of downstream task performance -- indicating the LLMs' probabilistic models of the world are misspecified. Lastly, we show how our measure can provide diagnostics to identify issues with LLM-powered inferential systems. --- ### Regulating Branch Parallelism in LLM Serving Authors: Swapnil Gandhi, Siva Hari, William J. Dally, Christos Kozyrakis Published: 2026-05-07 Categories: cs.DC, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2605.06914v1 Recent methods expose intra-request parallelism in LLM outputs, allowing independent branches to decode concurrently. Existing serving systems execute these branches eagerly or under fixed caps. We show that both are brittle: eager admission inflates the shared decode step, degrading co-batched requests in serial stages, while conservative fixed caps forgo the throughput that motivated exposing branches in the first place. We call the excess step latency caused by admitted branches the branch externality and show that the safe width depends on batch composition, context lengths, and accumulated slack, all of which change continuously over a workload trace. We introduce TAPER, a per-step admission controller that treats extra branches as opportunistic work, admitted only when the predicted branch externality fits within the batch's current slack budget. Per-step regulation is practical because branch-level scheduling decouples compute from memory: branches share the request's prefix KV, so expanding or contracting width requires no memory reclamation. On Qwen3-32B, TAPER improves goodput by $1.77\times$ over IRP-Off and by $1.48\times$ over IRP-Eager, while maintaining over $95\%$ SLO attainment. --- ### You Only Stack Once (YOSO): A Motion-Filtered, Deep-Learning Framework for Detecting Faint Moving Sources Authors: Nitya Pandey, César Fuentes, Pedro Bernardinelli, Valeria Frías, Colin Orion Chandler, David E. Trilling, Matthew J. Holman, Steven Stetzler, Dallin Spencer, Hsing Wen Lin, Luis E. Salazar Manzano, Darin Ragozzine et al. Published: 2026-05-07 Categories: astro-ph.EP, astro-ph.IM, cs.LG arXiv: https://arxiv.org/abs/2605.06913v1 We present You Only Stack Once (YOSO), an automated pipeline designed to detect faint, slow-moving Solar System objects in wide-field astronomical surveys. The pipeline integrates a novel Gaussian Motion Filter (GMoF) that operates at the pixel level to enhance signal-to-noise for objects exhibiting a range of apparent rates of motion. Unlike conventional shift-and-stack methods, which rely on discrete velocity trials, GMoF amplifies trails while suppressing random noise and static background features. Applied to a subset of DEEP observations from the Dark Energy Camera, YOSO recovered 45 out of 73 previously detected objects, as well as 11 new TNOs. It also discovered 216 objects in the near Solar System. Although alternative shift-and-stack methods are sensitive to objects about 0.88 magnitudes fainter, YOSO's false positive rate is extremely low, since it detects only sources that exhibit a trail and are consistent with a point source when shifted at the right rate. We show how this method can be deployed on large surveys like LSST, and adapted for other domains that require motion-based signal enhancement, including exoplanet imaging through Angular Differential Imaging (ADI), and near-Earth object (NEO) detection for missions like NEO Surveyor. YOSO thus provides a versatile, scalable approach for extracting faint, motion-dependent signals in the era of data-intensive astronomy. --- ### Advancing Reliable Synthetic Video Detection: Insights from the SAFE Challenge Authors: Kirill Trapeznikov, Gabriel Mancino-Ball, Jonathan Li, Paul Cummer, Jai Aslam, Danial Samadi Vahdati, Tai Nguyen, Matthew C. Stamm, Peter Bautista, Michael Davinroy, Laura Cassani, Jill Crisman Published: 2026-05-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.06912v1 The proliferation of generative video technologies has intensified the need for reliable methods to detect and characterize synthetic media. To address this challenge, we organized the \href{https://safe-video-2025.dsri.org}{SAFE: Synthetic Video Detection Challenge}, co-located with the \textit{Authenticity and Provenance in the Age of Generative AI (APAI) Workshop }at ICCV 2025. The competition invited participants to develop and evaluate algorithms capable of distinguishing real from synthetic videos under fully blind evaluation conditions with over 600 submissions from 12 teams over a 90 day span. Hosted on the Hugging Face platform, the challenge comprised two primary tasks: (1) detection of synthetic video content generated by diverse state-of-the-art models, and (2) detection of synthetic content following common post-processing operations such as resizing, re-compression, motion blur and others. The challenge data consisted of 13 modern high quality synthetic video models with generated content matched to real videos from 21 diverse and challenge sources, all adding up to 20 hours of 6,000 video samples. This paper describes the challenge design, dataset construction, evaluation methodology, and outcomes, offering insights into the generalization and robustness of contemporary synthetic video detection methods. Our findings highlight measurable progress in cross-generator generalization but also persistent vulnerabilities to post-processing artifacts. https://safe-video-2025.dsri.org --- ### Dual-Scale Temporal Fusion Reveals Structured Predictability in Subseasonal-to-Seasonal Temperature Prediction Authors: Elnaz Bashir, Jiali Wang, Lin Yan Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06911v1 Subseasonal-to-seasonal (S2S) temperature forecasts, spanning several weeks to a few months, are critically needed in agriculture practice, energy planning, and extreme-weather induced risk management, yet their reliability varies substantially across seasons and regions. Forecast skill is often attributed primarily to lead time, but this perspective does not fully explain the spatiotemporal patterns of predictability. Here we show that S2S predictability is organized across interacting temporal components, spatial heterogeneity, and large-scale pattern coherence, and that this structure can be explicitly characterized and exploited. We develop a dual-scale learning framework that separates calendar-aligned historical climate context from lead-time matched recent weather evolution, combining them through spatially adaptive fusion to enable stable temperature forecasts across the 30 to 90-day window. The learned fusion weights reveal that the balance between these two temporal scales shifts systematically with season and geography: during winter, interannual context dominates over high latitudes and complex terrain where forecast is the most difficult, while summer predictions reflect a more balanced temporal contribution across the domain. This spatially explicit reorganization of predictability, rather than simple lead-time decay, emerges as the primary determinant of forecast skill within the subseasonal window. Topology-aware structural constraints further improve spatial coherence of predicted temperature fields, stabilizing large-scale pattern organization particularly over complex terrain. These results reframe S2S predictability as a structured, multi-scale phenomenon, providing a more interpretable foundation for improving forecast systems and informing their use in practice. --- ### Same Signal, Opposite Meaning: Direction-Informed Adaptive Learning for LLM Agents Authors: Ziming Li, Jiatan Huang, Xiaoguang Guo, Guilin Wang, Chuxu Zhang Published: 2026-05-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.06908v1 Adaptive test-time compute for LLM agents aims to invoke extra computation only when it improves performance. Existing methods typically use confidence-, uncertainty-, or difficulty-based gates, assuming a fixed direction from the gating signal through compute need to the value of computation. This makes gating a utility-calibration problem: gating signals should align with whether extra computation improves the final outcome over the base policy. We show that this alignment is unstable: the same signal predicts rollout benefit in one setting and rollout harm in another, with reversals across environments and backbones even when the task is fixed. Wrong-direction gates can therefore worsen performance by precisely selecting harmful states. This reversal reflects a deeper distinction between compute need and compute suitability: a high uncertainty signal may indicate decision-difficult states where rollouts help compare alternatives, or intervention-unsuitable states where the current context does not support useful rollout-based improvement. Under this two-source model, fixed-direction gates are unreliable across heterogeneous settings. To address this, we propose DIAL (Direction-Informed Adaptive Learning), a sparse gate trained from signal-agnostic counterfactual exploration to learn the utility direction of state features per (environment, backbone). Across six environments and three backbones, DIAL yields a stronger overall success-cost trade-off than fixed-direction baselines. --- ### TraXion: Rethinking Pre-training Frameworks for Mobility and Beyond Authors: Shang-Ling Hsu, Mark Tenzer, Cyrus Shahabi, Khurram Shafique Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06906v1 Human mobility differs from text and from generic time series in three structural ways: visits are tuple-valued events whose meaning depends on the joint distribution over location, time, and activity; users carry persistent signatures across trajectories; and visits are not independent across users, since co-location at shared places is a primary signal. Existing pre-training recipes for mobility import objectives from language modeling, treating trajectories as sentences and visits as tokens, an analogy that fails against each of the three properties above. These properties define a broader class, multi-entity spatiotemporal event streams (MESES), spanning enterprise authentication logs, electronic health records, and other event-stream domains where entities share infrastructure, schedules, or contexts. We make the properties precise as three axioms that any pre-training framework for MESES should satisfy, and introduce TraXion, whose objectives and architecture are jointly designed to meet them. A single TraXion checkpoint per dataset beats task-specific baselines on every task across six public mobility datasets covering anomaly detection, next-POI recommendation, next-visit prediction, and social-link prediction. The same recipe, applied unchanged to enterprise authentication logs and ICU mortality prediction, matches or exceeds prior work on both, showing that event streams from domains as different as mobility, security, and healthcare can be modeled under a single framework. --- ### Conservative Flows: A New Paradigm of Generative Models Authors: Eshed Gal, Md Shahriar Rahim Siddiqui, Moshe Eliasof, Eldad Haber Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06905v1 Modern generative modeling is dominated by transport from a noise prior to data. We propose an alternative paradigm in which generation is performed by a discrete stochastic dynamics that leaves the data distribution invariant, initialized from data-supported states rather than from noise. The framework can utilize any pretrained flow model. We develop two probability-preserving sampling mechanisms, a corrected Langevin dynamics with a Metropolis adjustment and a predictor-corrector flow, that operate directly on existing checkpoints. We validate the framework on a synthetic Swiss-roll target, ImageNet-256 and Oxford Flowers-102, where our samplers consistently improve over the original generation procedures. --- ### MELD: Multi-Task Equilibrated Learning Detector for AI-Generated Text Authors: Chenjun Li, Cheng Wan, Haomiao Chen, Johannes C. Paetzold Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.06903v2 Large language models are widely used in everyday writing, making reliable AI-generated text detection crucial for academic integrity, content moderation, and provenance tracking. Yet high AUROC on clean, in-distribution benchmarks is not sufficient. Practical detectors must resist adversarial rewrites, generalize to unseen generators and writing domains, and maintain low false-positive rates (FPR). A pooled AI-versus-human objective does not explicitly require the model to distinguish among generator families, so it may fail to learn the generator-specific structure needed for generalization and attribution. We introduce MELD (Multi-Task Equilibrated Learning Detector), which instead trains on class-balanced, family-specific AI-versus-human tasks while sharing a common representation of human writing. MELD produces detection, generator-family, rewrite-task, and token-level predictions in a single forward pass. Format normalization further makes its scores invariant to the modeled reformatting operations. MELD ranks first among open-source submissions in the public RAID leaderboard and matches or exceeds supervised baselines on five of six held-out evaluation pools. To evaluate transfer to unseen generators, we introduce MELD-eval, a held-out test pool built from four frontier chat models. Without further fine-tuning, MELD achieves 99.7% TPR at 1% FPR on MELD-eval and 98% TPR at the same FPR on a held-out generator whose family is absent from training. Finally, in a case study of 5.9 million scientific texts from 2016--2026, MELD's prediction scores remain stable through 2022 and increase from 2023 onward, coinciding with the widespread adoption of LLM-based writing tools. The model, MELD-eval pool, source code, and live demo are available. --- ### MELD: Multi-Task Equilibrated Learning Detector for AI-Generated Text Authors: Chenjun Li, Cheng Wan, Johannes C. Paetzold Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.06903v1 Large language models are now embedded in everyday writing workflows, making reliable AI-generated text detection important for academic integrity, content moderation, and provenance tracking. In practice, however, a detector must do more than achieve high aggregate AUROC on clean, in-distribution human and AI text: it should remain robust to attacks and adversarial rewrites, transfer to unseen generators and domains, and operate at low false-positive rates (FPR). Most existing detectors optimize a single AI/Human objective, giving the representation little incentive to learn generator, attack, or domain structure once the binary task saturates. We introduce MELD (Multi-Task Equilibrated Learning Detector), a deployable detector for AI-generated text that enriches binary detection with auxiliary supervision. MELD attaches generator-family, attack-type, and source-domain heads to a shared encoder, and balances the four losses with learned homoscedastic uncertainty weights. To improve robustness, an EMA teacher predicts on clean inputs while an attack-augmented student is distilled toward the teacher. MELD further uses a hard-negative pairwise ranking loss to enlarge the score margin between AI-generated texts and the most confusable human texts. At inference, all auxiliary heads are discarded, giving MELD the same interface and cost as a standard detector. On the public RAID leaderboard, MELD is the strongest open-source detector and is competitive with leading commercial models, especially under attack and at low FPR. Across standard held-out benchmarks, MELD matches or outperforms supervised baselines. We further introduce MELD-eval, a held-out evaluation pool built from recent chat models released by four major LLM providers. Without additional finetuning, MELD achieves 99.9% TPR at 1% FPR on MELD-eval, while many baselines degrade sharply. --- ### Streaming Adversarial Robustness in Fuzzy ARTMAP: Mechanism-Aligned Evaluation, Progressive Training, and Interpretable Diagnostics Authors: Shane Cairns, Leonardo Enzo Brito da Silva, Sasha Petrenko, Donald C. Wunsch, Jian Liu Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06902v1 Adversarial robustness has been studied extensively for offline deep networks, but less is known about strict single-pass streaming neural learners. This paper studies adversarial robustness in Fuzzy ARTMAP, an Adaptive Resonance Theory architecture based on category competition, complement coding, match tracking, and replay-free prototype updates. We introduce WB-Softmax, a differentiable white-box attack surrogate aligned with ARTMAP's category-competition and map-field prediction mechanism, and formalize a streaming evaluation principle requiring robustness to be assessed on the final deployed model. Across four image benchmarks, WB-Softmax achieves 89-100% attack success on vanilla Fuzzy ARTMAP models. We show that defense rankings can reverse across protocols: offline adversarial training may appear strong under transfer attacks yet collapse under adaptive white-box evaluation, whereas progressive two-stage selective training provides the strongest overall replay-free robustness. We further show that ART's explicit category geometry enables interpretable diagnosis of separation collapse and match-score inversion. These results provide a mechanism-aligned, protocol-aware framework for adversarial robustness in streaming prototype-based learners. --- ### Reflections and New Directions for Human-Centered Large Language Models Authors: Caleb Ziems, Dora Zhao, Rose E. Wang, Matthew Jörke, Ahmad Rushdi, Advit Deepak, Sunny Yu, Anshika Agarwal, Harshvardhan Agarwal, Gabriela Aranguiz-Dias, Aditri Bhagirath, Justine Breuch et al. Published: 2026-05-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.06901v1 Large Language Models (LLMs) are increasingly shaping the private and professional lives of users, with numerous applications in business, education, finance, healthcare, law, and science. With this rise in global influence comes greater urgency to build, evaluate, and deploy these systems in a manner that prioritizes not only technical capabilities but also human priorities. This work presents a framework for developing Human-Centered Large Language Models (HCLLMs), which integrates perspectives from Natural Language Processing (NLP), Human-Computer Interaction (HCI), and responsible AI. Considering the ethics, economics, and technical objectives of language modeling, we argue that model developers need to address human concerns, preferences, values, and goals, not only during a cursory post-training stage, but rather with rigor and care at every stage of the pipeline. This paper offers human-centered insights and recommendations for developers at each stage, from system design to data sourcing, model training, evaluation, and responsible deployment. Then we conclude with a case study, applying these insights to understand the future of work with HCLLMs. --- ### Accelerated Relax-and-Round for Concave Coverage Problems Authors: Matthew Fahrbach, Mehraneh Liaee, Morteza Zadimoghaddam Published: 2026-05-07 Categories: cs.DS, cs.LG arXiv: https://arxiv.org/abs/2605.06900v1 We present an accelerated relax-and-round algorithm for concave coverage problems, which generalize the classic maximum coverage problem. Building on the relax-and-round framework of Barman et al. (STACS 2021), we propose two significant improvements. First, we replace the linear programming (LP) relaxation step with a projected accelerated gradient method applied to a smooth surrogate objective to achieve a $\widetilde{O}(mn \varepsilon^{-1})$ running time. Second, we use a specialized rounding scheme for the hypersimplex that combines the Carathéodory decomposition algorithm in Karalias et al. (NeurIPS 2025) with randomized swap rounding of Chekuri et al. (FOCS 2010). We prove tight approximation ratios for new reward functions, including a $0.827$-approximation for the logarithmic reward $\varphi(x) = \log(1 + x)$. Finally, we conduct maximum multi-coverage experiments on synthetic and real-world graphs, demonstrating that our algorithm outperforms approaches that use state-of-the-art LP solvers. --- ### Self-Programmed Execution for Language-Model Agents Authors: Luke J. O'Connor Published: 2026-05-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.06898v1 At the heart of existing language model agents is a fixed orchestrator program responsible for the state transition between consecutive turns. This paper introduces self-programmed execution (SPE), an agent architecture in which the model completion is itself the orchestrator program, and the harness evaluates this program but does not impose its own orchestration policy. I formalize this idea using agentic machines: an SPE state is one from which a model completion can load any state of an embedded copy of the machine, meaning that it is subject to no fixed turn-to-turn orchestration policy. Realizing SPE in practice is nontrivial because the same data is both model context and executable program. I therefore introduce Spell, a Lisp-based language in which programs can edit and re-evaluate themselves, and effectful expressions like model invocations are structured such that re-evaluating an edited program does not replay its side effects. Experiments with existing models, not trained for SPE or Spell, show that frontier models can operate in this regime and accomplish challenging agentic tasks. These results demonstrate how an LM can act as an agent without any fixed orchestration policy, and they raise the question of what self-orchestration strategies might be learned by a model trained for self-programmed execution. Code is available at https://github.com/lukejoconnor/spell . --- ### MIST: Multimodal Interactive Speech-based Tool-calling Conversational Assistants for Smart Homes Authors: Maximillian Chen, Xuanming Zhang, Michael Peng, Zhou Yu, Alexandros Papangelis, Yohan Jo Published: 2026-05-07 Categories: cs.CL, cs.AI, cs.HC, cs.MM, cs.SD, eess.AS arXiv: https://arxiv.org/abs/2605.06897v1 The rise of Internet of Things (IoT) devices in the physical world necessitates voice-based interfaces capable of handling complex user experiences. While modern Large Language Models (LLMs) already demonstrate strong tool-usage capabilities, modeling real-world IoT devices presents a difficult, understudied challenge which combines modeling spatiotemporal constraints with speech inputs, dynamic state tracking, and mixed-initiative interaction patterns. We introduce MIST (the Multimodal Interactive Speech-based Tool-calling Dataset), a synthetic multi-turn, voice-driven code generation task that operates over IoT devices. We find that there is a significant gap between open- and closed-weight multimodal LLMs on MIST, and that even frontier closed-weight LLMs have substantial headroom. We release MIST and an extensible data generation framework to build related datasets in order to facilitate research on mixed-initiative voice assistants which reason about physical world constraints. --- ### Mitigating Cognitive Bias in RLHF by Altering Rationality Authors: Tiffany Horter, Andrew Markham, Niki Trigoni, Serena Booth Published: 2026-05-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.06895v1 How can we make models robust to even imperfect human feedback? In reinforcement learning from human feedback (RLHF), human preferences over model outputs are used to train a reward model that assigns scalar values to responses. Because these rewards are inferred from pairwise comparisons, this learning depends on an assumed relationship between latent reward differences and observed preferences, typically modeled using a Boltzmann formulation in which a rationality parameter beta informs how consistently preferences reflect reward differences. In practice, beta is typically treated as a fixed constant that reflects assumed uniform annotator reliability. However, human feedback is not this simplistic in practice: real human judgments are shaped by cognitive biases, leading to systematic deviations from reward-consistent behavior that arise contextually. To address this, we treat rationality as context- and annotation-dependent. We design an approach to dynamically adjust the rationality parameter beta during reward learning using an LLM-as-judge to assess the likely presence of cognitive biases. This approach effectively downweights comparisons that are likely to reflect biased or unreliable judgments. Empirically, we show that this approach learns a more rational downstream model, even when finetuning on datasets with strongly biased preferences. --- ### McNdroid: A Longitudinal Multimodal Benchmark for Robust Drift Detection in Android Malware Authors: Md Mahmuduzzaman Kamol, Jesus Lopez, Saeefa Rubaiyet Nowmi, Emilia Rivas, Md Ahsanul Haque, Edward Raff, Aritran Piplai, Mohammad Saidur Rahman Published: 2026-05-07 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2605.06894v1 Machine learning (ML) in real-world systems must contend with concept drift, adversarial actors, and a spectrum of potential features with varying costs and benefits. Malware naturally exhibits all of these complexities, but for the same reason, it is challenging to curate and organize data to study these factors. We present McNdroid, to our knowledge the largest longitudinal multimodal Android malware benchmark for malware detection and drift analysis. McNdroid spans 2013--2025, excluding 2015, and represents each application with three aligned modalities--static features from manifests and smali code, dynamic behavioral features from sandbox execution, and graph-based features from function-call graphs. Using temporally separated splits, we evaluate standard ML and deep-learning detectors across increasing train--test time gaps. Results show clear temporal degradation, while multimodal fusion outperforms the best single modality across long-term temporal gaps. Cross-modal agreement also declines over time, suggesting that drift affects both individual feature spaces and the consistency among modalities. We further analyze modality-specific drift, malware-family evolution, and temporal changes in model explanations. We publicly release McNdroid, benchmark splits, and code to support reproducible research on temporal generalization and robust multimodal learning in security-critical, non-stationary settings. --- ### Not All Tokens Need 40 Steps: Heterogeneous Step Allocation in Diffusion Transformers for Efficient Video Generation Authors: Ernie Chu, Vishal M. Patel Published: 2026-05-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.06892v1 Diffusion Transformers (DiTs) have achieved state-of-the-art video generation quality, but they incur immense computational cost because standard inference applies the same number of denoising steps uniformly to every token in the sequence. It is well known that human vision ignores vast amounts of redundant motion. Why, then, do our densest models treat every spatiotemporal token with equal priority? In this paper, we introduce Heterogeneous Step Allocation (HSA), a training-free inference algorithm that assigns varying step budgets to different spatiotemporal tokens based on their velocity dynamics. To resolve the resulting sequence-length mismatch without sacrificing global context, HSA introduces a KV-cache synchronization mechanism that allows active tokens to attend to the full sequence while entirely bypassing inactive tokens. Furthermore, we derive a cached Euler update that advances the latent states of skipped tokens in a single operation without additional model evaluations. We evaluate HSA on the Wan-2 and LTX-2 models for both text-to-video (T2V) and image-to-video (I2V) generation. Our results demonstrate that HSA significantly outperforms previous state-of-the-art caching methods and the vanilla Flow Matching baseline, especially at aggressive acceleration regimes (e.g., 50% and 25% runtimes). Crucially, HSA achieves a superior quality-runtime Pareto frontier without the need for expensive offline profiling, robustly preserving structural integrity and generation quality even under tight computational budgets. Project page: https://ernestchu.github.io/hsa --- ### Towards Fairness under Label Bias in Image Segmentation: Impact, Measurement and Mitigation Authors: Aditya Parikh, Stella Frank, Sneha Das, Aasa Feragen Published: 2026-05-07 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2605.06891v2 Biases in annotation pipelines may introduce label bias: when annotation quality or style varies systematically across demographic subgroups, causing disparities in performance and auditability. Label bias in image segmentation remains underexplored, as even detecting it typically requires clean, unbiased annotations, which are not readily available. We present a Confident Learning framework adapted to segmentation for auditing label bias directly in the training data without a clean, unbiased ground truth. By comparing the provided training labels to the model's confident predictions, we measure the prevalence and direction of suspected label errors; where standard overlap metrics like Dice fail. We further show that label bias influences subgroup separability in the encoder's feature space, an artifact we leverage for bias mitigation rather than suppressing it. We evaluate three datasets spanning from synthetic to real-life bias and across experimental conditions, showing how our framework produces proxy audit signals without clean labels and mitigates disparities with a correctly identified cleaner reference group. --- ### Towards Fairness under Label Bias in Image Segmentation: Impact, Measurement and Mitigation Authors: Aditya Parikh, Stella Frank, Sneha Das, Aasa Feragen Published: 2026-05-07 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2605.06891v1 Labeled datasets reflect the biases of their annotation pipelines, which sometimes introduce label bias: group-conditional label errors that cause systematic performance disparities across demographic subgroups. Label bias in image segmentation remains underexplored, as even detecting it typically requires clean, unbiased annotations, which are not readily available. We present a data-centric adaptation of Confident Learning to segmentation, allowing detection of label bias directly in the training data without a clean, unbiased ground truth. By comparing the provided training labels to the model's confident predictions, we isolate directional errors that quantify the presence and nature of bias, where standard overlap metrics like Dice fail. We further show that label bias influences subgroup separability in the encoder's feature space, an artifact we leverage for bias mitigation rather than suppressing it. We evaluate three datasets, spanning from synthetic to real-life bias, showing how our framework reliably detects and mitigates bias without access to clean labels, achieving equitable performance across experimental conditions. --- ### Beyond the Black Box: Interpretability of Agentic AI Tool Use Authors: Hariom Tatsat, Ariye Shater Published: 2026-05-07 Categories: cs.AI, cs.MA arXiv: https://arxiv.org/abs/2605.06890v4 AI agents are promising for high-stakes enterprise workflows, but dependable deployment remains limited because these tool-use decisions are difficult to diagnose and control. Agents may skip required tool calls, invoke tools unnecessarily, or take actions whose consequences become visible only after execution. Existing observability methods are external: prompts reveal correlations, evaluations score outputs, and logs arrive only after the model has already acted. In long-horizon settings, these failures are costly because an early tool mistake can alter the rest of the execution trajectory, increase token consumption, and create downstream safety and security risk. We introduce a mechanistic-interpretability toolkit built on Sparse Autoencoders (SAEs), which decompose activations into sparse internal features, and linear probes, lightweight classifiers that read signals from those features. The framework reads model states before each action and infers whether a tool is needed and how risky the next tool action is. It identifies the model layers and features most associated with tool decisions and tests their functional importance through feature ablation. We train the probes on multi-step agent execution traces from the NVIDIA Nemotron function-calling dataset and apply the same workflow to GPT-OSS 20B and Gemma 3 27B models. The goal is not to replace external evaluation, but to add a missing layer: visibility into what the model signaled internally before action. This helps surface deeper causes of agent failure, especially in long-horizon runs where an early mistake can impact subsequent agent behavior. More broadly, the paper shows how mechanistic interpretability can support internal observability for monitoring tool calls and risk in agent systems. --- ### TriDE: Triangle-Consistent Translation Directions for Global Camera Pose Estimation Authors: Francisco Chen, Yiran Wang, Yunpeng Shi Published: 2026-05-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.06889v2 Pairwise translation directions are a key input to camera location estimation in global structure-from-motion. Existing estimators usually process each image pair independently, producing directions that may be locally plausible but inconsistent with the other relative directions in the viewing graph. To jointly estimate the direction, we propose TriDE, which exploits camera-triangle consistency as an efficient higher-order verification signal. Instead of solving a costly global nonlinear optimization problem that is sensitive to initialization, TriDE refines unreliable pairwise directions through message passing between directions and their incident weighted triangles. This information propagation strategy enables us to establish a strong phase-transition bound for exact recovery under a realistic random corruption model. Experiments on real image graphs show that TriDE improves direction accuracy by a large margin and yields better downstream camera locations, providing a practical link between local pairwise estimation and global camera pose geometry. --- ### TajPersLexon: A Tajik-Persian Lexical Resource and Hybrid Model for Cross-Script Low-Resource NLP Authors: Mullosharaf K. Arabov Published: 2026-05-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.06886v1 This work introduces TajPersLexon, a curated Tajik--Persian parallel lexical resource of 40,112 word and short-phrase pairs for cross-script lexical retrieval, transliteration, and alignment in low-resource settings. We conduct a comprehensive CPU-only benchmark comparing three methodological families: (i) a lightweight hybrid pipeline, (ii) neural sequence-to-sequence models, and (iii) retrieval methods. Our evaluation establishes that the task is essentially solvable, with neural and retrieval baselines achieving 98-99% top-1 accuracy. Crucially, we demonstrate that while large multilingual sentence transformers fail on this exact lexical matching, our interpretable hybrid model offers a favorable accuracy-efficiency trade-off for practical applications, achieving 96.4% accuracy in an OCR post-correction task. All experiments use fixed random seeds for full reproducibility. The dataset, code, and models will be publicly released. --- ### Don't Retrain, Align: Adapting Autoregressive LMs to Diffusion LMs via Representation Alignment Authors: Fred Zhangzhi Peng, Alexis Fox, Anru R. Zhang, Alexander Tong Published: 2026-05-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.06885v1 Diffusion language models (DLMs) have recently demonstrated capabilities that complement standard autoregressive (AR) models, particularly in non-sequential generation and bidirectional editing. Although recent work has shown that pretrained autoregressive checkpoints can be converted into diffusion language models, existing recipes primarily transfer parameters through continued denoising training with objective- and attention-level modifications. We instead ask whether the internal representation geometry learned by next-token prediction can be explicitly preserved during AR-to-DLM conversion. We hypothesize that much of the semantic structure learned by AR pretraining can transfer across generation orders, and thus DLM training should be viewed as relearning the decoding path rather than relearning language representations. To investigate this, we introduce REPR-ALIGN, a representation alignment objective that adapts a bidirectional masked diffusion model to reuse representations from a pretrained AR model of identical architecture. Concretely, we align the hidden states of the DLM to the frozen AR model at every layer using cosine similarity, while optimizing the standard masked denoising objective. This simple alignment, with no adapters and no architectural changes beyond the attention mask, yields up to 4x training acceleration in our setting and is particularly effective in low-data regimes. Our results suggest that linguistic representations can transfer across generation order, and that representation alignment provides a simple and effective technique for training diffusion language models. Code is available at https://github.com/pengzhangzhi/Open-dLLM. --- ### Muon with Nesterov Momentum: Heavy-Tailed Noise and (Randomized) Inexact Polar Decomposition Authors: Sayantan Choudhury, Xiaoran Cheng, Martin Takáč, Sen Na, Mladen Kolar Published: 2026-05-07 Categories: math.OC, cs.LG arXiv: https://arxiv.org/abs/2605.06884v1 Most first-order optimizers treat matrix-valued parameters as vectors, ignoring the intrinsic geometry of hidden-layer weights in neural networks. Muon addresses this mismatch by updating along the polar factor of a momentum matrix, but its theoretical understanding has lagged behind practice. In particular, practical implementations incorporate Nesterov momentum, compute the polar factor only approximately, and operate with stochastic gradients that may be heavy-tailed. We close this gap by developing a convergence theory for Muon with Nesterov momentum and inexact polar decomposition in non-convex matrix optimization under heavy-tailed noise. Our analysis builds on a unified framework for inexact polar decomposition that captures practical iterative approximations such as Newton-Schulz and quantifies how their errors propagate through the optimization dynamics. Under this framework, we establish an optimal iteration and sample complexity of $O \left(\varepsilon^{\frac{-(3α-2)}{(α-1)}} \right)$ for finding an $\varepsilon$-stationary point, where $α\in(1,2)$ denotes the heavy-tail index. For the inexact-polar setting with $σ_1=0$, we also provide guarantees that do not require prior knowledge of $α$. We analyze a randomized low-rank polar decomposition that is substantially more efficient than full-space methods while remaining compatible with our theory. Numerical experiments further demonstrate the effectiveness of the proposed inexact and randomized variants. --- ### Kernel Selection is Model Selection: A Unified Complexity-Penalized Approach for MMD Two-Sample Tests Authors: Yijin Ni, Xiaoming Huo Published: 2026-05-07 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2605.06883v1 The Maximum Mean Discrepancy (MMD) is a cornerstone statistic for nonparametric two-sample testing, but its test power is dictated entirely by the chosen kernel. Because any fixed kernel inherently fails to distinguish certain distributions, the kernel must be dynamically optimized. However, data-driven optimization violates the foundational i.i.d. assumption, forcing a strict trade-off in existing frameworks. Ratio criteria ignore this dependence, inducing overfitting and variance collapse on rich kernel classes. Conversely, aggregation methods bypass the dependence using finite grids, but this strategy cannot scale to continuous search spaces like deep kernels. To break this dichotomy, we establish data-driven kernel selection as a model selection problem. We propose Complexity-Penalized MMD (CP-MMD), a criterion derived by applying the two-sample uniform concentration inequality of preceding works to the post-optimization MMD problem. The resulting penalty bounds the empirical MMD by the complexity of the kernel search space, mathematically absorbing the cost of optimization, so that CP-MMD enables direct, grid-free maximization over continuous parametric classes, including scalar bandwidths, polynomial feature bandwidths, and deep network parameters. By formally accounting for optimization complexity, we prove that CP-MMD maximizes true test power while ensuring unconditional Type-I validity. Consequently, CP-MMD enables grid-free kernel selection across linear, polynomial-feature, and deep regimes, matching or exceeding state-of-the-art test power. --- ### How Well Do LLMs Perform on the Simplest Long-Chain Reasoning Tasks: An Empirical Study on the Equivalence Class Problem Authors: Chun Zheng, Lianlong Wu, Bingqian Li, Lvting Liu, Yi Zhou Published: 2026-05-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.06882v1 Large Language Models (LLMs) have achieved great improvements in recent years. Nevertheless, it still remains unclear how good LLMs are for reasoning tasks, especially for long-chain ones. In this paper, we evaluate LLMs' performance on the simplest yet long-chain reasoning task, namely the Equivalence Class Problem (ECP), i.e., determining whether two variables are equal given a set of randomly generated equivalence relations. We consider both reasoning and non-reasoning representative LLMs over a large variety of problem instances, ranging over different numbers of variables, connectivity probabilities, prompts, and other factors. The experimental results show that non-reasoning LLMs fail ECP, while reasoning models are significantly better but still struggle to completely solve this problem. Interestingly, considering various connectivity probabilities with a fixed number of variables, we observe that, for non-reasoning models, the hardest problem instances coincide with the phase transition point of ln n/(n-1), suggesting the chaos of the problem; in contrast, for reasoning models, the hardest ones coincide with the biggest diameter, suggesting the reasoning difficulty of the problem. --- ### Better Protein Function Prediction by Modeling Survivorship Bias Authors: Zhongmou Chao, Poompol Buathong, Ekaterina Selivanovitch, Susan Daniel, Peter I. Frazier Published: 2026-05-07 Categories: cs.LG, q-bio.QM arXiv: https://arxiv.org/abs/2605.06879v1 Protein sequence data from nature exhibits survivorship bias: we only observe data from those organisms that survive and reproduce, while non-functional protein mutations are eliminated by natural selection. Thus, predicting whether a protein sequence is functional often requires learning from positive examples alone. While positive-unlabeled (PU) learning frameworks offer a generic solution to this problem, existing PU methods ignore the evolutionary processes that shape sequence observability and cause survivorship bias. Consider a sequence that is one mutation away from a commonly-observed protein variant in a well-surveilled organism. If the sequence were functional, it would likely be observed. If it is not observed, this suggests non-functionality. In contrast, sequences that are unlikely to arise through mutation may be missing simply because they never arose. Thus, these two kinds of missing sequences should be treated differently when training models. In this work, we propose Evo-PU, a PU learning framework that uses a scientific understanding of nucleotide mutation to model survivorship bias for well-surveilled single-organism sequence data. On three prediction tasks using single-organism uniform-coverage surveillance data -- predicting results from held-out influenza and respiratory syncytial virus (RSV) mutagenesis studies, and predicting future SARS-CoV-2 variants -- Evo-PU outperforms standard PU learning, one-class classification (OCC), and protein language models (PLMs). On prediction tasks from multi-organism ProteinGym datasets with more heterogeneous surveillance coverage, we identify opportunities to generalize our approach. --- ### Temporal Attention for Adaptive Control of Euler-Lagrange Systems with Unobservable Memory Authors: Giansalvo Cirrincione, Adriano Fagiolini Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.06877v1 Adaptive control of Euler-Lagrange systems is challenging when friction is governed by a finite-horizon internal state that is not directly observable from joint measurements. In this setting, the measured closed-loop state is no longer Markovian, and standard certainty-equivalence adaptive laws may lose their convergence guarantees. The paper proposes a meta-control architecture in which the gains of a computed-torque controller are generated by a self-attention block processing a short window of recent motion history. The number of attention heads is selected before policy training through a surrogate analysis of the autocovariance of the memory-state gradient along the temporal window. This surrogate is based on a temporal adaptation of an incremental rank-tracking framework previously developed by the authors. The selected head count is then fixed and used as an architectural hyperparameter in a reinforcement-learning stage, where the policy is trained under a shielded admissibility constraint. The approach is tested on a 2-DOF manipulator with nonlinear friction and variable payload. In the short and matched memory regimes, the single-layer attention-only meta-controller outperforms a deeper Transformer baseline, with tracking-error reductions of 12 and 19 percentage points, respectively. The reported effect sizes are large, with d approximately -1.1 and -2.1, and Mann-Whitney p < 0.05 in both cases. In the long memory regime, however, the advantage disappears. Four out of ten training runs show either divergence or payload-invariant policy collapse, revealing a weakness in the static Phase-1 head-count prescription. This motivates moving rank-tracking inside the reinforcement-learning loop, allowing attention heads to be pruned or grown at runtime instead of fixed before training. --- ### AdpSplit: Error-Driven Adaptive Splitting for Faster Geometry Discovery in 3D Gaussian Splatting Authors: Yongjae Lee, Jingxing Li, Abhay Kumar Yadav, Rama Chellappa, Deliang Fan Published: 2026-05-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.06876v1 Adaptive density control in 3D Gaussian Splatting (3DGS) repeatedly grows the Gaussian population through fixed-cardinality random splitting to discover useful scene structure. However, in vanilla 3DGS, its binary split operator requires many densification rounds to expose fine details, making it a bottleneck for efficient training schedules with fewer iterations. We introduce AdpSplit, an error-driven adaptive split operator that determines the number of split children and initializes the child parameters from L1-pixel-error region statistics, enabling fewer densification iterations, thus reduced training time, while preserving the rendering quality of full-schedule training. Across the MipNeRF360, Deep-Blending, and Tanks&Temples datasets, AdpSplit reduces the training time of multiple accelerated 3DGS pipelines by 9.2%-22.3% as a simple drop-in replacement for the standard split operator. With FastGS, AdpSplit matches the full-schedule PSNR on MipNeRF360 while reducing training time by 16.4%, corresponding to a 12.6x acceleration over vanilla 3DGS. ---