# Research Papers — 2026-05 · bundle page 84 of 326 > Full-text markdown of papers #4151–4200 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/84 [← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/83.md) [Next bundle →](https://aipod.dev/papers/archive/2026-05/page/85.md) ### Perceive-then-Plan: Layout-as-Policy for Monocular 3D Scene Layout Estimation Authors: Junwei Zhou, Yu-Wing Tai Published: 2026-05-25 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.25326v1 Building structured 3D scene layouts from a single image requires reconciling visual observations with physical and spatial constraints, a challenge that is difficult to address with direct prediction alone. In this work, we formulate monocular 3D layout estimation as a perceive-then-plan problem with vision-language models, where a Perceiver first grounds the 3D objects and then a Planner iteratively refines the scene hypothesis through actions that improve physical plausibility while preserving consistency with the input image. We propose Layout-as-Policy (LaP), which casts the planning stage as a policy learning problem: 3D layouts are represented as structured states, and refined via discrete actions such as translation, rotation, and rescaling. Starting from an observation-aligned initialization with the geometry-enhanced Perceiver, the LaP Planner is trained to produce action sequences that progressively resolve geometric inconsistencies and enforce realistic spatial relations. To enable effective learning, we combine supervised trajectory initialization with preference-based optimization, allowing the model to learn corrective behaviors without requiring explicit reward engineering. This formulation transforms layout estimation from a one-shot prediction task into an iterative refinement process, enabling better handling of global constraints and complex object interactions. Experiments demonstrate that our approach produces layouts that are more physically coherent and better aligned with visual observations, while naturally supporting downstream tasks such as scene editing and manipulation. --- ### UWM-JEPA: Predictive World Models That Imagine in Belief Space Authors: Santosh Kumar Radha, Oktay Goktas Published: 2026-05-25 Categories: cs.LG, cs.AI, cs.RO, stat.ML arXiv: https://arxiv.org/abs/2605.25313v1 World models for partially observed environments must imagine multiple compatible hidden futures and steer between them under counterfactual actions. Joint Embedding Predictive Architectures (JEPAs) do this in latent space, but a vector-valued latent has no internal structure for carrying the belief over hidden continuations through blind rollout. We introduce the Unitary World Model JEPA (UWM-JEPA), a JEPA world model with a density-matrix latent on a joint system-environment space and a learned unitary predictor. The construction preserves the joint-state spectrum exactly during rollout, so the predictor itself cannot dissipate the represented uncertainty. On a hidden-velocity indicator task requiring five-step forward simulation under a given action sequence with the target observation masked, UWM-JEPA reaches 0.77 accuracy and degrades monotonically as actions are perturbed; a parameter-matched LSTM-JEPA trained under the same counterfactual-target objective and action head collapses to majority-class accuracy (0.53) under every action condition. Under blind rollout, UWM-JEPA loses fewer than ten points of probe R^2 at short horizons while vector-latent baselines lose forty-one and sixty-eight; both nevertheless tie on a held-out context probe, locating the separation in the predictor rather than the encoder. Action sensitivity itself requires training against counterfactual rather than teacher-forced targets, a finding that applies beyond the unitary parameterisation. For JEPA world models to imagine under partial observability, latent geometry and predictor dynamics matter, not frozen context-encoding capacity alone. --- ### P1SCO: Social Dimensions from a Perspectivist Lens Authors: Amanda Cercas Curry, Gianmarco de Francisci Morales, Luca Maria Aiello Published: 2026-05-25 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.25312v2 We introduce P1SCO, a dataset of social media comments collected from three distinct platforms, annotated according to ten social dimensions to capture the diversity of social interactions and perceptions. The dataset is carefully disaggregated to allow analysis at the level of individual comments, annotators, and platforms. In addition to the social dimension labels, we include rich metadata on the annotators, including demographics, Big Five personality profiles, and political affiliation. This combination of comment-level annotations and annotator-level features enables nuanced analyses of how social perception varies across platforms, individual differences, and demographic factors. By preserving the diversity of annotator perspectives, our dataset supports studies of inter- and intra-annotator agreement, the influence of personality and political orientation on social interpretation, and the cross-platform dynamics of social discourse. --- ### Tool-Call Dependency Structure is Linearly Decodable in LLM Agent Residual Streams Authors: Tianda Sun, Dimitar Kazakov Published: 2026-05-25 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.25310v1 Tool-using LLM agents produce trajectories whose calls form a directed dependency graph: earlier tool outputs supply arguments to later calls. Whether this execution structure is represented inside the model is unknown; prior structural probes have targeted static code or chain-of-thought text, not an agent's run-time call graph. A low-capacity edge probe on the residual stream of Qwen3-32B decodes the tool-call dependency graph well above both a Hewitt--Liang random-label control and a positional baseline. A counterfactual contrast between value corruption and structural perturbation indicates the signal tracks abstract topology rather than identifier values, and replicates under an independent, non-substring oracle. The non-positional component replicates on three further interactive multi-hop benchmarks and attenuates as call order alone becomes a sufficient proxy for dependency, vanishing in single-shot planning. Per-layer activation patching shifts the probe at a later, non-patched boundary, evidence that the representation propagates rather than passively reads out, though the realised tool call does not move. To our knowledge this is the first structural probe of an LLM agent's runtime tool-call dependency graph. Our claims concern representation, not behavioural control, and span two model families and one primary domain. --- ### Stabilizing Streaming Video Geometry via Dynamic Feature Normalization Authors: Xiaoyang Lyu, Muxin Liu, Xiaoshan Wu, Ruicheng Wang, Yi-Hua Huang, Yang-Tian Sun, Shaoshuai Shi, Xiaojuan Qi Published: 2026-05-25 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.25308v1 Consistent 3D geometry estimation from streaming RGB input is crucial for real-world applications such as autonomous driving, embodied AI, and large-scale reconstruction. While modern monocular geometry foundation models achieve strong single-image accuracy, they exhibit severe temporal inconsistency on continuous input, notably dominated by scale--shift drifting. Through targeted empirical analysis, we trace this instability to its root cause: fluctuations in latent feature statistics, whose mean and variance directly determine the predicted depth's scale and shift. Building on this insight, we introduce Dynamic Feature Normalization (DyFN), a lightweight, causal recurrent module that dynamically and robustly modulates feature statistics to maintain stable geometry over time. We adapt powerful pretrained monocular geometry models for streaming by finetuning only DyFN, a mere 2\% additional parameters, while keeping the backbone frozen, thereby achieving temporal consistency without compromising single-image accuracy. Extensive experiments across four benchmarks show that DyFN effectively eliminates temporal artifacts such as disjointed layering and positional jitter, and achieves state-of-the-art temporal stability, improving over prior streaming methods by up to 14\% and even outperforming heavier non-causal video baselines. Project Page: https://shawlyu.github.io/DyFN --- ### Recursive Class Connectivity Classification (R3C) Applied to Binary Image Segmentation for Improved Infant Fingerprint Enhancement Authors: Joao Leonardo Harres Dall Agnol, Luiz Fernando Puttow Southier, Jefferson Tales 0liva, Marcelo Teixeira, Rodrigo Mineto, Marcelo Filipa, Dalcimar Casanova, Erick Oliveira Rodrigues Published: 2026-05-25 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.25307v1 Image enhancement plays a crucial role in infant fingerprint matching, as child-specific characteristics such as smaller finger dimensions and thinner ridge structures often degrade image quality during acquisition. To address these limitations, enrollment typically depends on specialized highresolution scanners, which most existing enhancement methods are not designed to support. Consequently, identification rates for children remain significantly lower than those achieved with adult fingerprints. This study introduces Recursive Class Connectivity Classification (R3C), a novel framework that iteratively refines binary segmentation outputs from existing enhancement methods by extending ridge structures. R3C does not require modifications to the underlying classifier and operates without training data, which is not currently available for infant fingerprints. Instead, the method improves segmentation by repeatedly feeding the classified image back into the classification process, while combining each intermediate segmentation with the original input image. Experiments conducted on three fingerprint datasets using four different enhancement classifiers show that R3C can increase the True Acceptance Rate (TAR) by up to 4% for children and over 40% for newborns, compared to using the enhancement methods alone. A qualitative analysis further demonstrates that R3C reconnects fragmented ridge patterns, improving the visual quality of segmentation. Because it functions independently of the enhancement method used, R3C provides a flexible and broadly applicable solution for improving binary segmentation. --- ### Electricity Consumption Forecasting: An Approach Using Cooperative Ensemble Learning with SHapley Additive exPlanations Authors: Eduardo Luiz Alba, Gilson Adamczuk Oliveira, Matheus Henrique Dal Molin Ribeiro, Érick Oliveira Rodrigues Published: 2026-05-25 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.25305v1 Electricity expense management presents significant challenges, as this resource is susceptible to various influencing factors. In universities, the demand for this resource is rapidly growing with institutional expansion and has a significant environmental impact. In this study, the machine learning models long short-term memory (LSTM), random forest (RF), support vector regression (SVR), and extreme gradient boosting (XGBoost) were trained with historical consumption data from the Federal Institute of Paraná (IFPR) over the last seven years and climatic variables to forecast electricity consumption 12 months ahead. Datasets from two campuses were adopted. To improve model performance, feature selection was performed using Shapley additive explanations (SHAP), and hyperparameter optimization was carried out using genetic algorithm (GA) and particle swarm optimization (PSO). The results indicate that the proposed cooperative ensemble learning approach named Weaker Separator Booster (WSB) exhibited the best performance for datasets. Specifically, it achieved an sMAPE of 13.90% and MAE of 1990.87 kWh for the IFPR-Palmas Campus and an sMAPE of 18.72% and MAE of 465.02 kWh for the Coronel Vivida Campus. The SHAP analysis revealed distinct feature importance patterns across the two IFPR campuses. A commonality that emerged was the strong influence of lagged time-series values and a minimal influence of climatic variables. --- ### When Interpretability Becomes a Liability: Adversarial Attacks on CBM Concept Layers Authors: Aditya Sridhar Published: 2026-05-25 Categories: cs.LG, cs.CR, cs.CV arXiv: https://arxiv.org/abs/2605.25304v1 Concept Bottleneck Models (CBMs) have emerged as a cornerstone approach for interpretable machine learning, providing human-understandable intermediate representations through explicit concept activations. However, this interpretability fundamentally introduces a critical, previously unexplored attack surface: the concept bottleneck layer itself. We present a comprehensive, systematic study of concept-level adversarial vulnerabilities in CBMs, revealing that targeted, minimal perturbations operating on input pixels can induce catastrophic misclassification by manipulating semantic representations. We develop a rigorous theoretical framework to quantify concept-space robustness, establishing novel metrics that expose the vulnerability landscape of these architectures. Our extensive analysis on the CUB-200-2011 dataset demonstrates that standard CBMs exhibit severe susceptibility to concept-level manipulation. To address this critical weakness, we introduce SPECTRA (Semantic Perturbation-based Concept Training for Robustness against Attacks), a principled stability regularization defense. SPECTRA effectively hardens the semantic representation space, increasing the minimal perturbation norm required for a successful attack from 0.46 to over 4,200, rendering targeted concept manipulation computationally prohibitive. Furthermore, SPECTRA preserves baseline classification accuracy to within 2.2%. By establishing concept-level attacks as a fundamentally distinct threat model, this work opens a new research frontier at the intersection of interpretable machine learning and adversarial robustness. --- ### Seeing Without Understanding: Large Language Model Evaluation of Mobile User Interface Quality, Failure Taxonomy, and Architectural Explanation Authors: Md Rejaul Korim Sadi, Golam Mostofa Naeem, Toufiqur Rahman Tasin, Syed Mostofa Moosa, Mahmudul Hasan Emon, Mahmudur Rashid, Ferdus Ahmed Published: 2026-05-24 Categories: cs.HC, cs.CL, cs.SE arXiv: https://arxiv.org/abs/2609.05423v1 Evaluating mobile user interface quality at scale remains a persistent challenge in software engineering and human-computer interaction. Rule-based heuristic methods offer structural reliability but demand significant engineering effort, while human annotation does not scale to the volume of applications produced annually. Large language models present a promising alternative, yet their reliability for structured UI judgment has not been systematically examined, and the patterns behind their failures remain insufficiently characterized. This paper addresses both gaps. We begin with the complete RICO dataset of 66,261 real-world mobile application screens, from which we derive a refined evaluation corpus of 15,000 screens through a rigorous, literature-guided selection process. Each screen is assessed across seven criteria: structural JSON validity, minimum visible element count, clickable component presence, non-zero layout bounds, image integrity, and perceptual duplicate removal. Against this corpus, we apply a heuristic baseline built from severity-weighted usability signals, normalized layout metrics, and pixel-ratio complexity measures calibrated to real user sentiment. Multiple language models independently rate each screen across usability, layout quality, and visual complexity from structured JSON descriptions and raw screenshots. Dimension-level comparison against the heuristic uses agreement rates, Cohen's Kappa, and confidence calibration. Recurring divergence patterns are organized into a failure taxonomy and interpreted through transformer architectural signatures: MLE plausibility bias, attention misgrounding, and autoregressive over-commitment. --- ### Exploring the Role of LLMs in HPC Programming: A Survey Authors: Strahinja Ljaljevic, Josep Jorba, Sergio Iserte Published: 2026-05-24 Categories: cs.DC, cs.AI arXiv: https://arxiv.org/abs/2608.26110v1 Large Language Models (LLMs) are emerging as promising assistants in High-Performance Computing (HPC), where programming remains complex and expertise-intensive. This survey systematically reviews their application across five categories: code generation, parallelization and optimization, frameworks and architectures, evaluation and benchmarking, and broader challenges. The analysis highlights both opportunities and limitations: while general-purpose LLMs perform reasonably well on serial and OpenMP-like tasks, they fall short in distributed paradigms such as MPI, where correctness and scalability are critical. Domain-specialized models (e.g., HPC-Coder, HPC-GPT, chatHPC) achieve higher accuracy through fine-tuning, curated datasets, and retrieval-augmented generation (RAG), yet their scope remains narrow and their evaluations largely limited to benchmarks or micro-kernels. The broader picture is one of dual potential and fragility: LLMs can lower barriers to entry, accelerate prototyping, and support code modernization, but they remain brittle under production-level requirements where correctness, performance portability, and scaling cannot be compromised. We conclude that LLMs are unlikely to replace HPC experts in the near term but are positioned to become powerful collaborators in the software development pipeline. Their effective deployment will require richer datasets, integration with performance analysis and schedulers, rigorous evaluation frameworks, and governance structures that ensure transparency and trust. The convergence of AI and HPC should therefore be understood as a long-term, co-evolutionary process, where each advance uncovers new challenges and opportunities for reshaping scientific software development. --- ### Safe Evolution with Circuit Anchors Authors: Yan Liu, Jie Fu, Tsung-Yi Ho Published: 2026-05-24 Categories: cs.CL, cs.LG, cs.NE arXiv: https://arxiv.org/abs/2608.05158v1 In biological evolution, unconstrained mutation can lead to catastrophic outcomes: organisms may evolve enhanced capabilities while losing essential functions for survival. Nature's solution is \textit{developmental constraints}, where core regulatory genes remain anchored while peripheral genes adapt freely. We observe that current self-evolution algorithms for large language models lack analogous constraints. They optimize purely for capability, implicitly assuming safety will be preserved. Our experiments reveal this assumption to be dangerously wrong: models can \textit{misevolve} into powerful yet dangerous entities. Inspired by how Hox genes anchor body structure across $500$ million years of evolution, we propose \textbf{Circuit-Anchored Evolution (CAE)}. Using mechanistic interpretability, we identify a tiny \textit{safety circuit}, comprising less than $2$\% of model features, that causally mediates safety behaviors. We anchor this circuit during evolution, constraining it within a small displacement bound while allowing the remaining features to evolve freely. This mirrors the biological principle of \textit{evolvability with constraint}: preserving what is essential while adapting what is peripheral. Experiments across $3$ model families and two evolution algorithms demonstrate that CAE achieves superior safety preservation with minimal capability loss, substantially outperforming explicit reward-based constraints in both effectiveness and efficiency. Just as developmental constraints prevent biological evolution from producing nonviable organisms, circuit anchoring prevents model evolution from producing capable but dangerous systems. --- ### SonicSampler: Unified Tile-Aware Kernels for LLM Sampling and Speculative Verification Authors: Pragaash Ponnusamy, Shivam Sahni, Jue Wang, Tri Dao Published: 2026-05-24 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2607.20475v1 Sampling in LLM inference comprises a combinatorial set of logit processing, token selection, and verification operations for speculative decoding. However, existing implementations either accelerate only subsets of this pipeline, rely on multiple kernel launches, or assume homogeneous sampling behavior across a batch, limiting support for dynamic serving workloads and preventing efficient CUDA Graph execution. We present $\textbf{SonicSampler}$, a unified suite of tile-aware Triton kernels that vertically fuses the complete sampling pipeline into a fixed, workload-aware execution model. Our kernels support dynamic per-request sampling behaviors, including grammar-constrained decoding, repetition, frequency and presence penalties, logit bias, temperature scaling, top-$k$ / top-$p$ / min-$p$ filtering, and speculative verification - within a single batched kernel while remaining fully CUDA Graph-compatible. Central to our approach is a novel hierarchical two-stage top-$k$ algorithm that achieves up to $\textbf{10x speedup}$ over competitive baselines and exploits the low-entropy structure of LLM outputs to enable efficient selection over large vocabularies. Across heterogeneous speculative decoding workloads, SonicSampler achieves up to $\textbf{16x speedup}$ over state-of-the-art baselines while preserving flexible batched execution. --- ### VeriSimpl: Robust Optimization Modeling from Natural Language using Simplification-based Verification Authors: Sumaya Abdul Rahman, Seckhen Ariel Andrade Cuellar, Ghani Raissov, Mohammad Raza Published: 2026-05-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.20474v2 Natural language interfaces can greatly benefit the accessibility and usability of optimization modeling, and recent advances in large language models (LLMs) show promise in automatically translating textual problem descriptions into executable solver formulations. However, a key challenge for existing approaches is to ensure that the inferred formulation correctly implements the intended task, even if it may execute without errors. We introduce VeriSimpl, a solver LLM framework for robust natural-language-to-optimization formalization. Our approach is based on the idea of simplification-based verification, where the optimization solver is leveraged to generate simplified diagnostic queries about a candidate formulation to allow the LLM to tractably reason about the correctness of the formulation with respect to the task description. We present such simplification strategies along different dimensions with respect to problem constraints and decision variables, which allow the LLM to reason locally under fixed global contexts. Evaluations on a range of optimization benchmarks show how our approach provides consistent improvements in accuracy over existing methods, while also providing a novel high-precision self-verification signal. --- ### VeriSimpl: Robust Optimization Modeling from Natural Language using Simplification-based Verification Authors: Sumaya Abdul Rahman, Seckhen Ariel Andrade Cuellar, Ghani Raissov, Mohammad Raza Published: 2026-05-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.20474v1 Natural language interfaces can greatly benefit the accessibility and usability of optimization modeling, and recent advances in large language models (LLMs) show promise in automatically translating textual problem descriptions into executable solver formulations. However, a key challenge for existing approaches is to ensure that the inferred formulation correctly implements the intended task, even if it may execute without errors. We introduce VeriSimpl, a solver LLM framework for robust natural-language-to-optimization formalization. Our approach is based on the idea of simplification-based verification, where the optimization solver is leveraged to generate simplified diagnostic queries about a candidate formulation to allow the LLM to tractably reason about the correctness of the formulation with respect to the task description. We present such simplification strategies along different dimensions with respect to problem constraints and decision variables, which allow the LLM to reason locally under fixed global contexts. Evaluations on a range of optimization benchmarks show how our approach provides consistent improvements in accuracy over existing methods, while also providing a novel high-precision self-verification signal. --- ### Incomplete Prompt Jailbreaks in Large Language Models Authors: Yeonjea Kim, Bumjin Park, Jaesik Choi Published: 2026-05-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.20473v1 Large language models (LLMs) are increasingly released as open-weight models with safeguards against harmful requests. Nevertheless, sentence completion remains vulnerable to incomplete harmful prompts. In this work, we formalize this phenomenon as incomplete prompt jailbreaks (IPJ) and provide a systematic empirical characterization of when and how incomplete prompts elicit harmful continuations. We analyze diverse attractor types associated with incomplete sentence continuation and show that LLMs systematically delay refusal until sentence termination. We further demonstrate that training models to refuse incomplete harmful prompts via parameter tuning is insufficient, failing to generalize across both content domains and attractor types. To enable fine-grained control, we identify two functional neurons: termination and continuation neurons. By clarifying their roles in sentence completion, we highlight the potential of neuron-level interventions for more precise and robust IPJ defenses. --- ### Robust Critics: Defending LLMs Against Multi-Turn Attacks Authors: Roman Belaire, Arunesh Sinha, Pradeep Varakantham Published: 2026-05-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.20472v1 When a user asks a language model something harmful, is it a genuine attack or a misunderstood but well-meaning question? This ambiguity is one of the central challenges of LLM safety. A model that assumes the worst harms legitimate users; one that assumes the best is easily exploited. The problem is compounded in multi-turn dialogue, where an attacker's true intent may only reveal itself gradually across many exchanges, yet existing safety frameworks apply a contextual bandit treatment, ignoring the trajectory of the conversation. To that end, we propose Dialogue Critic Guided Sampling (DCGS), a framework that addresses this by inferring user intent at every turn of dialogue. Instead of applying a fixed rule about what is or is not safe, DCGS learns what the user's intent is likely to be based on the full conversational history and generates responses accordingly. Formally, we model adversarial dialogue as a Markov Decision Process and learn value and regret-based critics at both the individual token and utterance (full response) levels, scoring candidate responses via an action-value critic. We prove that this inference-time reweighting approximates exponential tilting of the base policy, guaranteeing improvement in expected return for any finite candidate pool, a property that group-relative objectives do not exhibit. Evaluated on CARES-18k, WildJailbreak, Redbench, and Harmbench, DCGS outperforms strong robust baselines and frontier models on adversarial dialogue tasks. DCGS also transfers to frontier models, improving their robustness without fine-tuning. --- ### Catalyst Papers in Artificial Intelligence Research: A Landscape on ICLR from 2017 to 2025 Authors: Fan Huang Published: 2026-05-24 Categories: cs.DL, cs.AI arXiv: https://arxiv.org/abs/2607.05401v1 A small number of methodological contributions, including word2vec, the Transformer, large-scale pre-training, and reinforcement learning from human feedback, have reshaped NLP and AI research over the past decade. OpenReview now makes numeric reviewer scores and accept/reject decisions public for every ICLR submission. Whether such review signals identify trajectory-changing papers at submission time, however, remains untested at corpus scale. We answer this question on $36{,}113$ papers from ICLR 2017--2025, identifying \emph{catalysts}: papers whose descendants measurably redirect future research. We compare four disruptiveness measures (the Consolidation/Destabilization (CD) index, node2vec, the direction-aware Embedding Disruptiveness Measure (EDM), and an LLM-based semantic rater) and define a five-type operational catalyst taxonomy (topic initiator, topic bridge, within-topic redirector, simultaneous, and recognition-misaligned). EDM leads at identifying highly cited ICLR papers (AUC $0.83$ vs.\ $0.60$ for CD, $0.49$ for node2vec, and $0.42$ for the LLM rater). Topic initiators precede a $7.55{\times}$ topic-share growth and topic bridges precede an $11.52{\times}$ growth in cross-topic citation flow versus year-matched controls. We found that the peer review scores are essentially orthogonal to future disruptiveness ($|ρ|{\leq}0.005$; accepted and rejected papers have indistinguishable mean EDM, $p{=}0.11$). --- ### OverFlowLight: Real-Time Gridlock Prevention and Traffic Signal Optimization for Urban Intersections Authors: Mingyuan Li, Boyang Huang, Tianqi Jiang, Chenpu Li, Chunyu Liu, Yang Li, Ruimin Li, Qiang Wu Published: 2026-05-24 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2606.27381v1 Queue overflow, a severe consequence of urban traffic congestion, occurs when vehicle queues exceed intersection capacity, obstructing upstream traffic and triggering cascading gridlocks. Prevailing traffic signal control (TSC) algorithms, primarily optimized for throughput, often fail to address overflow during peak hours, exacerbating congestion and creating safety hazards. We propose OverFlowLight, a real-time framework designed to preemptively resolve overflow and enhance overall TSC performance. It first introduces a mechanism to accurately detect overflow in real-time by leveraging multi-modal sensing from cameras and radars. Upon detection, it dynamically generates and inserts dedicated overflow phases into the signal cycle to clear the blocking queues. This is orchestrated by a hybrid control design that combines rapid rule-based overflow intervention with controller back ends such as reinforcement learning (RL) for longer-horizon efficiency. We conducted extensive real-world deployments of OverFlowLight across 43 intersections in three major cities. The framework demonstrates seamless integration with existing RL-based TSC agents, highlighting its modularity and practical applicability. Empirical results show that OverFlowLight reduces overflow incidents by 60.4% and increases network throughput by 18.2% compared to deployed baselines. Furthermore, it substantially diminishes the need for manual intervention common with expert-tuned signal plans. This work presents the first practical, scalable, and data-driven framework for actively preventing traffic gridlock, offering a crucial component for building resilient and efficient urban transportation systems. Our demonstration videos, codes and datasets are available at the anonymous URL, https://anonymous.4open.science/r/OverFlowLight-FBF9. --- ### Analysis and Prediction of At-Risk Students Using Machine Learning Algorithms Authors: Soheila Gheisari, Hamid Salarian Published: 2026-05-24 Categories: cs.CY, cs.LG arXiv: https://arxiv.org/abs/2606.20617v1 Student attrition represents a significant challenge for higher education institutions because it impacts both academic results and financial viability. Machine learning provides an effective solution to identify students who require assistance before they leave their academic programs. The research investigates how machine learning approaches enable institutions to predict student withdrawal and enrollment cancellation through data-driven insights for strategic decisionmaking. The evaluation of models includes Logistic Regression, Random Forest, Support Vector Machines (SVM), and K-Nearest Neighbors (KNN) based on academic performance and demographic data and enrollment records. The results show that logistic regression and linear SVM models produced the highest accuracy which demonstrates ML's capability to detect students at risk. --- ### Specifying AI-SDLC Processes: A Protocol Language for Human-Agent Boundaries Authors: Ylli Prifti, Pasquale De Meo, Alessandro Provetti Published: 2026-05-24 Categories: cs.AI, cs.MA, cs.PL, cs.SE arXiv: https://arxiv.org/abs/2606.20615v3 AI agents now act as first-class members of the software development lifecycle, but the instruments teams use to direct them enforce nothing: process encoded in prompts is flexible but unenforceable, while workflow formalisms are enforceable but do not model autonomous agents. We propose a domain-specific language for specifying AI-SDLC processes as protocols, with formal abstract syntax, well-formedness conditions, operational semantics, and enforcement invariants, organised around a separation of policy (declared intent) from mechanism (structural enforcement). We prove that any well-formed protocol maintains its invariants on every execution trace, closed under Kleene composition of orchestration loops: protocol steps cannot be skipped, by construction. A failure-rate analysis derives two consequences: structural enforcement bounds silent failure while converting the remainder into visible, audited stalls, and the benefit carries a capability floor. We validate the design in simulation and end-to-end on SWE-bench Verified, where an identical bug-fix methodology yields no improvement delivered as prose instructions, but a significant, replicated 14-22 point gain when executed as a validated, self-correcting process; an ablation attributes the gain to the executed process itself, and the capability floor appears where the model predicts it. As foundation models converge, the durable engineering asset is the formally specified, executable process. --- ### Specifying AI-SDLC Processes: A Protocol Language for Human-Agent Boundaries Authors: Ylli Prifti Published: 2026-05-24 Categories: cs.AI, cs.MA, cs.PL, cs.SE arXiv: https://arxiv.org/abs/2606.20615v2 AI agents now participate as first-class team members across the software development lifecycle, yet no specification language exists for expressing the human-agent responsibility boundaries, approval gates, and governance constraints this collaboration requires. Existing approaches encode process in agent prompts (subject to drift), target adjacent domains (workflow management, business processes), or address only fragments (access control, approval gates). We propose a domain-specific language for specifying AI-SDLC processes as protocols, with formal syntax, well-formedness conditions, operational semantics, and enforcement invariants. The language distinguishes policy (declared intent) from mechanism (structural enforcement), enabling implementations to bound process non-determinism through primitives such as validation tokens and capability boundaries. Three results follow. A failure rate analysis shows that structural enforcement bounds system failure rates at a weighted product of agent and validator rates, while behavioral compliance permits cumulative or near-saturating growth. The 2+N team pattern (two human-in-control roles plus N specialized agent members) formalizes classical Separation of Duties for AI-SDLC. Kleene closure of orchestration loops and reflexive protocol-adherence validation emerge as design properties rather than special-case constructs. We position the contribution against multi-agent frameworks (MetaGPT), workflow specification (FlowAgent, BPMN extensions), and capability-based security (SAGA): the novelty lies in the specific integration, not any single primitive. A working implementation demonstrates feasibility; empirical evaluation is future work. --- ### A Geometric Gaussian Mixture Representation of Plane Curves Authors: Ali Darijani, Benedikt Stratmann, Jürgen Beyerer Published: 2026-05-24 Categories: cs.CG, cs.AI, cs.CV, math.DG arXiv: https://arxiv.org/abs/2606.06505v1 We introduce a user defined probabilistic polygonal representation for plane curves. Given a curve, we select vertices on the curve and connect consecutive vertices by line segments to obtain a polygonal approximation. Each segment is equipped with a user defined uncertainty parameter in the normal direction. This yields a collection of thin probabilistic geometric primitives that retain the geometrz of the underlying curve while extending it beyond the idealized deterministic one dimensional formulation. For each segment, we define a Random Variable that is uniform distributed in the tangent direction of the segment and Gaussian distributed in the normal direction of the segment. By matching the first and the second central moments, this construction induces a Gaussian component whose mean lies at the segment midpoint and whose covariance encodes both tangential and normal uncertainty. Combining the segment wise components with appropriate weights yields a Gaussian Mixture Model (GMM) representation of the user defined probabilistic polygonal representation of the plane curve. The proposed framework provides an analytically tractable probabilistic model that preserves local geometry, and uncertainty in the normal direction. It applies to smooth, closed, open, non regular, and self intersecting plane curves, allows adaptive discretization and varying uncertainty in the normal direction, and as a result supports uncertainty aware geometric modeling. Experiments on a collection of canonical plane curves show that the resulting GMM capture local tangent, local normal, and local arc length; resulting in the global shape of the underlying curves to be truthfully captured as well. The representation is particularly relevant for applications in uncertainty aware CAD and digital twins, probabilistic obstacle modeling in robotics, and probabilistic trajectory planning. --- ### Samudra 2: Scaling Ocean Emulators across Resolutions Authors: Yuan Yuan, Jesse Rusak, Alexander Merose, Adam Subel, Pavel Perezhogin, Alistair Adcroft, Carlos Fernandez-Granda, Laure Zanna Published: 2026-05-24 Categories: cs.CE, cs.AI, cs.LG, physics.ao-ph arXiv: https://arxiv.org/abs/2606.02610v2 Ocean general circulation models (OGCMs) are essential to climate science but computationally expensive, limiting ensemble size and forcing scenarios. Neural emulators promise orders-of-magnitude speedups, yet existing ocean emulators have not combined fine spatial resolution with multi-year autoregressive rollouts. Samudra, the first autoregressive neural ocean emulator to produce multi-decade global rollouts, is limited to $1^\circ$ resolution and exhibits two long-horizon failure modes: \emph{variance collapse}, the loss of temporal variability, and \emph{imprinting artifacts}, in which velocity patterns leak into deep-ocean fields. We present Samudra 2, which introduces a wider U-Net backbone with modified ConvNeXt-style blocks and a reduced block-internal expansion factor, together with a dynamic loss that reweights output channels according to their prediction errors, strengthening gradients for slow-evolving deep-ocean fields. At $1^\circ$, Samudra 2 increases upper-ocean global-mean temperature $R^2$ from 0.56 to 0.87 and reduces deep-ocean temperature error by roughly sevenfold. The same architecture scales to $1/2^\circ$ and $1/4^\circ$ over approximately 8-year autoregressive rollouts, recovering mesoscale eddies and sharp western boundary currents. Running on a single GPU, Samudra 2 enables larger ensembles for sea-level projections, ocean heat uptake, and climate variability studies. All artifacts are publicly available: project page, code, checkpoints, documentation. --- ### Aligning Cellular Sheaves with Classifier Attention for Interpretable Weakly-Supervised Pathology Localization Authors: Devansh Lalwani, Swapnil Bhat, Maulik Shah Published: 2026-05-24 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2606.00092v1 Weakly-supervised classification of whole-slide images with attention-based multiple instance learning (ABMIL) on top of foundation features now reaches near-saturation on Camelyon16 slide-level performance, but the corresponding attention maps are an imperfect localization signal: in clinical interpretation, a model that classifies correctly without firing on the actual lesion is hard to trust. We address this gap with cellular sheaves, which equip each vertex and edge of a graph with a finite-dimensional vector space and consistent linear maps between them, providing a principled way to detect local disagreement on graph-structured data. We apply cellular sheaves to weakly-supervised tumour localization on whole-slide images, combining a sheaf disagreement field with ABMIL. The natural training objective, encouraging consistency between similar features, produces a disagreement field that tracks tissue-level texture rather than diagnostic content. We propose attention-conditional consistency, which uses the classifier's attention to define which neighbouring patches should agree. Joint training of the classifier and the sheaf under this objective produces a disagreement field with patch-level AUC 0.940 on Camelyon16 and raises the attention head from its ABMIL-alone level of 0.717 to 0.953. Two-stage ablation with the classifier frozen at its ABMIL values reaches only 0.727 on the disagreement field and leaves attention at 0.717, confirming that the gain comes from the projector co-adapting under both objectives, not from the loss change in isolation. The trained model transfers without retraining to annotated slides from Camelyon17, maintaining Delta AUC 0.932 +/- 0.083 and attention AUC 0.955 +/- 0.099. The result is an attention map and a sheaf-disagreement map that fire on the same diagnostic regions, giving clinicians two complementary explanations for each slide-level prediction. --- ### DLLM-JEPA: Joint Embedding Predictive Architectures for Masked Diffusion Language Models Authors: Sangdae Nam Published: 2026-05-24 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2606.00091v1 Joint Embedding Predictive Architectures (JEPAs) have reshaped self-supervised representation learning in vision. The recent LLM-JEPA ported JEPA to autoregressive language models but inherited two steep costs from the causal-attention substrate: it demands explicit multi-view data (e.g., text-code pairs), and it requires two gradient-carrying forward passes per step. We introduce DLLM-JEPA, which pairs JEPA with masked-diffusion language models to eliminate both costs at once. The bidirectional attention of diffusion models yields two semantically distinct views of the same input via different masking rates -- no explicit pairs needed -- and supports a single gradient-carrying forward pass, cutting training FLOPs by 33% relative to LLM-JEPA. DLLM-JEPA improves over diffusion-only fine-tuning in every (task, architecture) combination we evaluate: up to +18.7 pp on LLaDA-8B GSM8K and +11.4 pp on Dream-7B GSM8K, with consistent positive gains on Spider, NL-RX-SYNTH, and Django. Beyond accuracy, DLLM-JEPA exhibits a dual-win property: on LLaDA-8B with the Wide-t configuration, it simultaneously raises GSM8K accuracy (67.1 vs. 65.2, +1.8 pp), drives held-out Wikitext loss below the pre-trained base, and preserves MMLU accuracy at base level across three fine-tuning seeds -- whereas an L2-to-base parameter anchor matches baseline accuracy with no task gain. Layer-wise probing reveals the mechanism: a geometric-functional drift dissociation in which the fine-tuned backbone moves further from the pre-trained weights than the baseline yet forgets less on held-out Wikitext, with the amplification concentrated in middle transformer layers. The pattern appears on Dream-7B as well, indicating the phenomenon is not specific to a single backbone. --- ### Gait2Hip-60: A Unified Deep Learning Benchmark for Predicting Hip Muscle Forces and Joint Moments from Multi-Cadence Gait Kinematics Authors: Jiaqi Zhang, Ji Hou, Qing Sun, Xianzhi Gao, Bo Huo Published: 2026-05-24 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.30374v1 Estimating hip muscle forces and joint moments during gait typically relies on musculoskeletal simulation, which is informative but time-consuming and difficult to apply in clinical settings. This study developed a deep learning framework to predict these hip dynamics parameters directly from lower-limb gait kinematics and compared three representative sequence models under a unified protocol. Gait data were collected from 60 healthy adults under three metronome-guided cadence conditions. Ten bilateral lower-limb joint angles were used as inputs, and OpenSim-derived hip muscle forces and hip joint moments were used as reference outputs. Three deep learning models of LSTM, Transformer, and Mamba were trained and evaluated using the same subject-level split, preprocessing pipeline, and metrics. The best model was then directly tested on an external cohort of 9 patients with osteonecrosis of the femoral head (ONFH) without retraining. In the healthy-subject benchmark, Transformer achieved the best subject-level mean performance for both hip muscle force prediction (RMSE = 1.33 N/kg, MAE = 0.57 N/kg, R2 = 0.819) and hip joint moment prediction (RMSE = 0.11 Nm/kg, MAE = 0.07 Nm/kg, R2 = 0.862), with similar advantages across walking cadences. In zero-shot external validation, Transformer retained moderate predictive ability in ONFH for hip muscle force prediction (RMSE = 1.51 N/kg, MAE = 0.70 N/kg, R2 = 0.537) and hip joint moment prediction (RMSE = 0.17 Nm/kg, MAE = 0.12 Nm/kg, R2 = 0.569). These findings support the feasibility of estimating hip dynamics from gait kinematics, identify Transformer as a strong baseline, and highlight the need for broader pathological validation and improved generalization before clinical application. --- ### Fine-Tuning Vision-Language Models for Understanding Current Damage and Scoring Priority with Quality Guard Agent Authors: Takato Yasuno Published: 2026-05-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.27452v1 Bridge inspection in Japan requires mandatory visual assessments every five years, yet qualitative damage ratings (levels a-e) assigned by different engineers exhibit significant inter-rater variability -- a critical barrier to consistent infrastructure management. The aging of skilled engineers further threatens inspection capacity. This paper presents a methodology for automating bridge damage understanding and repair priority scoring using fine-tuned Vision-Language Models (VLMs). We fine-tune LLaVA-1.5-7B with QLoRA on up to 4,000 paired bridge damage images and inspection text records, then evaluate on a fixed test set of 800 images. The model outputs natural language descriptions identifying structural members and damage patterns, from which a rule-based scoring engine calculates a five-level repair priority index. A progressive training study (1k/2k/3k/4k samples) reveals that 2k training samples achieve near-optimal validation loss in only 2.9 hours of training; beyond 2k, validation loss improves by no more than 0.2% per doubling of training samples, exhibiting clear diminishing returns. Furthermore, semantic similarity on the held-out test set peaks at 3k (0.6909) and degrades at 4k (0.6739), indicating that quality-curated mid-scale data outperforms larger but noisier corpora. Inference optimization combining torch.compile() and batch processing (batch_size=8) achieves 10.06 seconds per image -- a 70.2% reduction over the unoptimized baseline. Our approach contributes to data governance in bridge inspection, reduces inter-rater variability, and provides AI-assisted triage to augment expert engineers in inspection workflows. Furthermore, we introduce a two-stage Quality Guard using a fine-tuned Swallow-8B SLM to reject low-quality VLM outputs before priority scoring, preventing spurious scores from damaged or unrecognised images. --- ### From Affect to Complex Behavior: Advancing Multimodal Human-Centered AI at the 10th ABAW Workshop & Competition Authors: Dimitrios Kollias, Panagiotis Tzirakis, Alan Cowen, Stefanos Zafeiriou, Irene Kotsia, Eric Granger, Marco Pedersoli, Simon Bacon, Jens Madsen, Soufiane Belharbi, Muhammad Haseeb Aslam, Chunchang Shao et al. Published: 2026-05-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.27451v1 The 10th Affective & Behavior Analysis in-the-Wild (ABAW) Workshop and Competition, held at CVPR 2026, continues to advance research on modelling, analysis, understanding of human affect and behavior in real-world, unconstrained environments. The workshop maintains its dual structure, comprising both a competition and a paper track. The ABAW Competition introduces a diverse set of challenges targeting key aspects of affective and behavioral understanding, including continuous affect (valence-arousal) estimation, discrete affect (expression and action unit) recognition, as well as more complex behavior analysis tasks, such as emotional mimicry intensity estimation, ambivalence/hesitancy recognition and fine-grained violence detection. These challenges are built upon large-scale in-the-wild datasets, providing comprehensive benchmarks for state-of-the-art approaches. In parallel, the paper track presents a wide range of contributions spanning pose, motion & behavior estimation, affect modelling & multimodal learning, benchmarks, datasets & evaluation protocols, fairness, robustness & deployment. Overall, the 10th ABAW Workshop and Competition continues to serve as a key platform for benchmarking, collaboration and innovation, shaping the development of next-generation multimodal, human-centered AI systems. --- ### Context Features Are Cheap: Rank-Aware Decomposition for Efficient Feature Interaction in Recommender Systems Authors: Yevgeny Tkach Published: 2026-05-24 Categories: cs.IR, cs.LG arXiv: https://arxiv.org/abs/2605.27450v1 Modern industrial recommender systems use a deep ranking model to score N candidates against the same user and context features. Standard implementations broadcast context features early in the forward pass, redundantly computing context-only operations N times per request. We present a rank-aware decomposition applicable to the dominant interaction mechanisms in modern recommender architectures-Factorization Machine (FM) pairwise products, Deep Cross Network (DCNv2) cross layers, self-attention, and fully connected (FC) projection layers-built on a single algebraic principle: any linear or bilinear operation over a rank-partitioned input admits an exact block decomposition that moves context-only computation from once-per-candidate to once-per-request, identity-equivalent to the original model. Closed-form analysis and controlled ablation verify that savings scale quadratically with the number of context features. Applied to a production DLRM-style ranker without any architectural change, the decomposition increases per-pod throughput by 87.5% (a 47% reduction in peak pod count) at identical model predictions. The identity-equivalent decomposition applies only at the first layer of cross networks and self-attention, since each layer mixes ranks in its output. To extend savings across depth, we further introduce rDCN, an architectural variant of DCNv2 that maintains rank discipline across depth and matches DCNv2 accuracy within training noise at 67% fewer total FLOPs, and sketch an analogous architectural variant for self-attention. --- ### Checking Fact with Better Retrieval: Dynamic Contrastive Learning for Evidence Retrieval Authors: Zhongtian Hua, Yi Luo, Meijia Yu, Yingjie Han Published: 2026-05-24 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2605.27449v1 In the field of multimodal fact checking, the accuracy of retrieving evidence from different modalities has a significant impact on the downstream claim verification process. Existing general multimodal retrieval methods are often constructed based on semantics, resulting in the retrieved evidence being similar but not relevant to the claim. This paper proposes a \textbf{D}ynamic \textbf{A}daptive \textbf{C}ontrastive \textbf{L}earning method for evidence \textbf{R}etrieval called DACLR to address these issues. DACLR first uses a Multimodal Large Language Model (MLLM) to uniformly convert multimodal evidence and claims into text modalities, and extracts the features of these information at event level. Then, it conducts evidence retrieval through a two-stage retrieval method of recall-rerank. DACLR enhances the model's event perception ability of the retrieval stage by optimizing the contrastive loss and mining hard negative samples. Specifically, DACLR designs three loss functions at two levels (semantic and event) based on the InfoNCE loss.Corresponding to these, three sets of hard negative sample candidates are set up. The model dynamically adjusts the ratio based on the accuracy supervision signal of intra-batch samples, allowing the model to learn the correlation between claims and positive samples at the event level without forgetting the semantic retrieval ability. Extensive comparison and ablation experiments demonstrates the effectiveness of DACLR and its internal optimization methods. Further research also prove the advantages of DACLR in the field of multimodal evidence retrieval. --- ### Planning Neural Dynamics with Lie Group Embedding through Supervised Projective Manifold Learning Authors: Tianwei Wang, Bryan Chen, Qian Zuo, Qiyue Xia, Xin Li, Wei Pang Published: 2026-05-24 Categories: cs.LG, cs.AI, math.DS, math.RA arXiv: https://arxiv.org/abs/2605.26167v1 We propose Lie group embedded dynamical neural networks (LieEDNN) and the corresponding learning algorithms based on gradient descent and metric projection on smooth manifold, where we treat Lie group as an intrinsic representation for continuous symmetry of manifold geometry. Thereby we achieve learnable and stable dynamics on the underlying manifold for general Lie group, and we are able to utilize the powerful representation capability of Lie group such as SO(3) and SE(3) to solve real world engineering problems in areas such as robotics, graphics, and control. Two core challenges are: (i) General Lie groups are incompatible with addition arithmetic, which is necessary for neural network interactions. (ii) The dynamics evolve in the nonlinear representation space of special algebra rather than the normal Euclidean space, which violates the paradigm of common neural ODEs. To address these two challenges, we firstly introduce adjoint Lie group action on the Lie algebra, which induces a linear mapping and transfer to the block-wise structure of weight matrices, such that addition could operate on the Lie algebra as a vector space. Then we parameterize the Lie algebra and the adjoint action as linear transformation so that the architecture is aligned with neural network perceptrons. Explicitly, this embedding appears as block-wise manifold constraints on weights, and we develop algorithms to learn the equilibrium with stability guarantees of the temporal neural network dynamics. Experiments are implemented on a specific Lie group SE(3), with the application scenario of telescopic manipulators. --- ### Enhancing Autonomous Online Intrusion Detection for IoT with Balanced Learning, Reliable Pseudo-Labels, and Lightweight Architectures Authors: Hanzala Afzaal, Danish Memon, Chouhdary Bilal Raza, Muhammad Khurram Shahzad Published: 2026-05-24 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.26166v1 The rapid proliferation of Internet of Things (IoT) devices has created an urgent demand for adaptive, resource-efficient Intrusion Detection Systems (IDS) capable of handling dynamic and evolving cyber threats. This paper investigates AOC-IDS, a state-of-the-art autonomous online IDS published at IEEE INFOCOM 2024, which employs an Autoencoder (AE) with Cluster Repelling Contrastive (CRC) loss and an autonomous Gaussian-based decision module. We first successfully replicate AOC-IDS on the UNSW-NB15 benchmark, achieving 89.39% accuracy in close agreement with the published 89.19%. We then identify four key limitations: class imbalance, unreliable pseudo-label generation, limited generalization, and computational overhead for IoT deployment, and propose targeted improvements for each. Our XGBoost-BalSamp method achieves 95.45% accuracy on UNSW-NB15, a gain of 6.26% over the baseline. Our combined deep learning approach (PseudoFilter, MixupAug, and LiteAE) achieves a best-run accuracy of 90.88% (F1: 91.45%), surpassing the base paper while reducing model parameters by 55%.These results demonstrate that targeted improvements to AOC-IDS yield consistent accuracy gains while improving practical deployability on IoT edge devices. --- ### Tool-Schema Compression Enables Agentic RAG Under Constrained Context Budgets Authors: Furkan Sakizli Published: 2026-05-24 Categories: cs.SE, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2605.26165v1 Agentic RAG systems that equip language models with dozens to hundreds of tool definitions face a critical resource conflict: tool schemas consume the same context window needed for retrieval-augmented generation. We present the first systematic study of this tool-context trade-off, evaluating 14 models spanning 1.5B-32B local models plus one frontier API model across 6,566 controlled API calls at three context budgets (8K, 16K, 32K) with 28 tool definitions. Applying TSCG conservative-profile compression (44-50% schema token savings), we observe a binary enablement effect: at 8K tokens, JSON-schema tool definitions overflow the context window entirely, yielding near-zero EM (2.6% average), while compressed schemas restore RAG functionality with +20.5 pp average exact-match lift across all eight models (+24.7 pp among the six exhibiting full enablement). At 32K -- where both formats fit -- four of five tested models show delta <= 1 pp, confirming the effect is purely budget-driven. External validation on HotpotQA (50 multi-hop questions) shows +48 pp EM under the same overflow scenario. Frontier scaling tests demonstrate that JSON schemas overflow at ~494 tools while compressed schemas remain operational beyond 800 tools. Our results establish tool-schema compression as a necessary infrastructure layer for agentic RAG in constrained-context deployments. All code, data, and checkpoints are publicly available. --- ### Adversarial Water-Filling: Theory, Algorithms, and a Domain-Specific Wireless Foundation Model Authors: Xindi Tong, Chee Wei Tan, H. Vincent Poor Published: 2026-05-24 Categories: cs.IT, cs.LG, math.OC arXiv: https://arxiv.org/abs/2605.26163v2 Competitive resource allocation problems over frequency and space can be formulated as minimax interaction between transmit power and worst-case interference. This formulation naturally arises in multi-operator low Earth orbit (LEO) satellite spectrum sharing, where transmissions from competing constellations interfere in real-time. Under Gaussian channels, the corresponding power-allocation problem admits a convex-concave formulation with a unique saddle point. Discrete constellations yield generally nonconvex mercury/water-filling formulations. In this paper we propose the adversarial water-filling (AWF) problem with corresponding theory and algorithms for these settings. In addition, we develop a domain-specific wireless foundation model for AWF to learn the AWF search dynamics. The architecture incorporates permutation-invariant channel representations, a constraint-aware graph neural network (GNN) with sparse message passing, and global latent variables capturing the low-dimensional water level implied by the AWF optimality. Through learned projected extragradient iterations, the model approximates stationary solutions of the constrained minimax problem arising under mercury/water-filling. We further establish projected-stationarity/Karush-Kuhn-Tucker consistency and conditional local convergence of the learned AWF dynamics under local regularity and stability conditions. Experiments demonstrate empirical generalization across unseen problem sizes, constraint structures, discrete constellations, and channel-weighted objectives, while achieving a median speedup exceeding one order of magnitude over Mirror-Prox on matched instances at comparable first-order solution quality. The related code can be found at https://github.com/convexsoft/AWF. --- ### Adversarial Water-Filling: Theory, Algorithms and Foundation Model Authors: Xindi Tong, Chee Wei Tan, H. Vincent Poor Published: 2026-05-24 Categories: cs.IT, cs.LG, math.OC arXiv: https://arxiv.org/abs/2605.26163v1 Competitive resource allocation problems over frequency and space can be formulated as minimax interaction between transmit power and worst-case interference. This formulation naturally arises in multi-operator low Earth orbit (LEO) satellite spectrum sharing, where transmissions from competing constellations interfere in real-time. Under Gaussian channels, AWF is strongly convex--concave on nondegenerate active channels, whereas discrete constellations yield generally nonconvex mercury/water-filling formulations. In this paper we propose the Adversarial Water-Filling (AWF) problem with corresponding theory and algorithms for these real situations. In addition, we develop a wireless foundation model for AWF to learn the AWF search dynamics. The architecture incorporates permutation-invariant channel representations, a constraint-aware graph neural network (GNN) with sparse message passing, and global latent variables capturing the low-dimensional water level implied by the AWF optimality. Through learned projected extragradient iterations, the model approximates stationary solutions of the constrained minimax problem arising under mercury/water-filling. We further show that, under local regularity and contractivity conditions, the learned AWF dynamics converge locally linearly around regular stationary points. Experiments demonstrate empirical generalization across unseen problem sizes, different constraints, and multiple discrete constellations, while achieving more than one-order-of-magnitude runtime improvements over iterative baselines. The related code can be found at https://github.com/convexsoft/AWF. --- ### On the Push-Based Asynchronous Federated Learning: A Bias-Correction Aggregation Approach Authors: Jiahui Bai, Hai Dong, A. K. Qin Published: 2026-05-24 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.26162v1 Asynchronous decentralized federated learning (ADFL) eliminates central coordination and global synchronization, making it attractive for large-scale and heterogeneous systems. However, frequent peer-to-peer communication, asynchronous updates on directed topologies, and non-IID data jointly lead to excessive communication overhead, biased aggregation and severe model drift. We propose PushCen-ADFL, a communication-efficient ADFL framework that enables stable training under asymmetric communication and delayed client participation. PushCen-ADFL couples communication, aggregation, and local stabilization in a shared centroid representation space, forming a closed loop between compression and optimization. Clients exchange centroid-form messages, apply average-preserving push-sum mixing to correct aggregation bias, and use a lightweight centroid regularization anchored in the same centroid space to mitigate drift under heterogeneity and staleness. A bounded, sender-deduplicated buffer further improves robustness under irregular asynchronous arrivals. Experiments on vision datasets demonstrate that PushCen-ADFL improves accuracy under data heterogeneity by up to 6\% while reducing per-push communication cost by more than 80\%, achieving a favorable accuracy-communication trade-off. --- ### TSFMAudit: Data Contamination Auditing in Forecasting Time Series Foundation Models Authors: Hongkai Li, Shifeng Xie, Lefei Shen, Zhuo Li, Mouxiang Chen, Xiaobin Zhang, Han Fu, Jianling Sun, Xiaoxue Ren, Chenghao Liu Published: 2026-05-24 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.26161v1 Time series foundation models (TSFMs) are increasingly pretrained on large corpora, raising concerns that evaluation datasets may have been exposed during pretraining and thus yield overly optimistic performance estimates. Auditing such contamination is challenging in time series because signals are continuous and heterogeneous, and often lack corpus documentation. To the best of our knowledge, this is the first work to study pretraining contamination auditing for TSFMs. We formalize the problem of pretraining contamination auditing for TSFMs and propose TSFMAudit, a method based on probe adaptation dynamics. Our key intuition is that contamination manifests as unusually efficient adaptation: after a fine tuning probe, contaminated datasets tend to exhibit faster loss reduction with smaller backbone movement. We evaluate TSFMAudit on 6 TSFMs and 187 datasets using documented training source evidence as supervision, and compare against 10 competitive baselines adapted from the LLM literature. --- ### Device Context Protocol: A Compact, Safety-First Architecture for LLM-Driven Control of Constrained Devices Authors: Dongxu Yang Published: 2026-05-24 Categories: cs.NI, cs.CR, cs.LG arXiv: https://arxiv.org/abs/2605.26159v1 Large language models are increasingly used as orchestrators of external tools via the Model Context Protocol (MCP), but MCP is built for software services with megabytes of memory and does not descend to the microcontrollers that dominate the long tail of physical devices. Recent work (IoT-MCP) ports MCP to edge gateways at 74 KB peak memory; this still excludes the smallest commodity MCUs and, critically, does not address the safety problem of giving an unreliable caller (an LLM that may hallucinate or be prompt-injected) direct control of physical hardware. We present the Device Context Protocol (DCP): a sub-50-byte typical frame (6-byte header + CBOR payload + optional 16-byte HMAC), a manifest schema in which capability scoping, range and type checks, dry-run evaluation, and units-as-types are protocol-layer primitives, and a host-side Bridge that rejects malformed or hallucinated calls before any byte reaches the device. Reference firmware measures 27.6 KB flash / 0.6 KB RAM on ESP32; the Python Bridge, ESP32 firmware, and a language-neutral conformance suite are MIT-licensed and public. An empirical study -- 675 tool calls produced by five LLMs across four vendors (DeepSeek, Alibaba, Zhipu, MiniMax) against six categories of adversarial prompts, with the injection category instantiating AgentDojo's attack templates -- shows DCP rejects 100% of capability-escalation attempts and 78% of prompt-injection attempts, versus 0--1% for Raw MCP and IoT-MCP, matching the expressiveness of a well-formed OpenAPI 3 schema at three orders of magnitude less firmware footprint. We position DCP as the missing layer between MCP (which is moving toward enterprise SaaS connectivity) and the physical devices it does not reach. --- ### Furina: Fragmented Uncertainty-Driven Refusal Instability Attack Authors: Tongxi Wu, Jian Zhang, Yang Gao Published: 2026-05-24 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.26158v1 Safety alignment in large language models (LLMs) and multimodal large language models (MLLMs) is commonly assumed to operate as a near-binary threshold mechanism. We challenge this assumption by revealing that safety behavior is governed by an instability region where small perturbations induce stochastic refusal decisions rather than deterministic outcomes. We develop a multi-metric diagnostic framework combining external and internal signals to characterize this instability. Through systematic experiments, we identify a characteristic diagnostic signature: inputs in unstable regimes exhibit elevated output uncertainty yet decreased internal safety activation, a decoupling phenomenon that explains why detection-based defenses fail against sophisticated attacks. Building on this framework, we introduce Furina, a jailbreak attack that deliberately induces this signature through fragmented, scene-anchored prompts without model-specific optimization. Furina outperforms strong single-turn and multi-turn baselines on HarmBench and achieves competitive results on MM-SafetyBench, demonstrating that uncertainty amplification provides a principled and transferable mechanism for understanding safety vulnerabilities. Code is available at: https://github.com/0xCavaliers/Furina_Jailbreak. --- ### Turning Bias into Bugs: Bandit-Guided Style Manipulation Attacks on LLM Judges Authors: Xianglin Yang, Bryan Hooi, Gelei Deng, Tianwei Zhang, Jin Song Dong Published: 2026-05-24 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.26156v2 The known stylistic biases in LLM judges, such as a preference for verbosity or specific sentence structures, present an underexplored security vulnerability. In this work, we introduce BITE (BIas exploraTion and Exploitation), a black-box adversarial framework that learns semantics-preserving edits to mislead an LLM judge and artificially inflate the scores it assigns. We cast the selection of stylistic edits as a contextual bandit problem and use a LinUCB policy to adaptively choose edits that maximize the judge's score without access to model parameters or gradients. Empirically, we test BITE across a diverse range of LLM judges and tasks, including both pointwise and pairwise comparisons on chatbot leaderboards and AI-reviewer benchmarks. BITE achieves an attack success rate exceeding 65% and raises scores by 1-2 points on a 9-point scale, all while preserving semantic equivalence. We further assess the attack's stealthiness, showing that BITE evades standard style-control methods and several detection baselines. Our findings expose a fundamental weakness in the LLM-as-a-judge paradigm and motivate robust, attack-aware evaluation. Our code is available at https://github.com/xianglinyang/llm-as-a-judge-attack. --- ### When Does Adaptive Guidance Help? Belief-Aware Privileged Distillation for Autonomous Driving Under Partial Observability Authors: Mehmet Haklidir Published: 2026-05-24 Categories: cs.RO, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.26155v1 Guided Soft Actor-Critic (GSAC) distills knowledge from a privileged full-state teacher to a partial-observation student for autonomous driving, but uses a fixed distillation coefficient lambda regardless of the agent's uncertainty. We present Belief-Aware GSAC (BA-GSAC), which modulates lambda via ensemble disagreement, and use it as a testbed for a systematic empirical study asking: when does adaptive guidance actually help? Evaluating five strategies (fixed lambda in {0.01, 0.1}, adaptive, linear decay, and vanilla SAC) across three POMDP difficulty levels on Highway-Env, we find that preliminary single-seed runs suggest benefits under mild and moderate partial observability, but under severe occlusion (evaluated with 3 seeds for all methods) the adaptive coefficient collapses to lambda_min within about 3K steps. We trace this to an observability blindness phenomenon: because the ensemble predicts partial observations, it achieves low disagreement even under heavy occlusion, modeling what is visible but unable to detect what is missing. We diagnose the root cause and propose an architectural fix (training the ensemble on full-state predictions using the guiding actor's privileged access); while not validated here, we show that even with current limitations, the warmup phase provides measurable stabilization (CV=13.3% vs. 29.8% for constant lambda=0.01). In fact, a simple deterministic linear decay schedule achieves the best severe-POMDP performance across all metrics (mean 116.5, CV=8.9%), suggesting that the scheduling effect, not the ensemble, drives the stability benefit. These findings provide practical guidance for designing uncertainty-aware teacher-student frameworks and highlight ensemble prediction targets as an important design choice. --- ### MemMorph: Tool Hijacking in LLM Agents via Memory Poisoning Authors: Xuanye Zhang, Yongsen Zheng, Zhuqin Xu, Kaiyu Zhou, Bowen Shen, Haoran Ou, Tianwei Zhang, Kwok-Yan Lam Published: 2026-05-24 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2605.26154v1 LLM-driven agents are capable of selecting external tools to complete users' tasks. However, attackers could compromise such process, steering agents toward inappropriate/wrong tools and enabling malicious actions. Most existing attacks primarily manipulate the tool metadata, which is easily detectable by auditing and may lose effectiveness as modern agents increasingly adopt memory modules to refine tool selection policies through accumulated experience. This paper proposes MemMorph, the first attack that bias tool selection by poisoning the agent's long-term memory. Rather than explicitly dictating the tool invocation decision, MemMorph injects a small number of crafted records that are disguised as technical facts, incident reports, and operational policies. These poisoned records reshape the agent's contextual perception and decision-making process, leading it to autonomously infer and select the tool preferred by the attacker. Experiments across 3 benchmarks, 10 agent backbones, and 3 memory-module implementations show that MemMorph achieves up to 85.9% attack success rate with only three injected records, outperforming the strongest baseline by up to 25% while retaining potency under 3 representative defenses. Our findings expose long-term memory as a critical and under-explored attack surface in tool-augmented agents, urging the development of memory-level integrity safeguards. --- ### Algorithms with Polynomially-Improved Approximation Factors for the $2 \rightarrow q$ Norm, and Applications Authors: Samuel B. Hopkins, Stefan Tiegel Published: 2026-05-24 Categories: cs.DS, cs.LG, math.ST, stat.ML arXiv: https://arxiv.org/abs/2605.25303v3 The $2 \rightarrow q$ norm of a matrix $X \in \mathbb{R}^{n \times d}$ is defined as $\lVert X \rVert_{2 \rightarrow q} = \sup_{\lVert v \rVert_2 = 1} \lVert Xv \rVert_q$. We give polynomial-time multiplicative approximation algorithms for this norm when $q > 2$ (i.e. in the hypercontractive setting). This problem either directly captures or is closely related to long-standing open problems in combinatorial optimization and hardness of approximation (e.g. Small Set Expansion), quantum information (e.g. Best Separable State), and algorithmic statistics. Very little is known about what approximation factors we can achieve for this problem in polynomial time, even though such approximations have significant downstream consequences. Barak, Brandão, Harrow, Kelner, Steurer, and Zhou showed that no polynomial-time algorithm can achieve an approximation factor better than $2^{\sqrt{\log n}}$, assuming the Exponential Time Hypothesis (FOCS'12). On the other hand, a simple spectral algorithm gives a $d^{1/4}$-approximation as a baseline. For the important special case of $q = 4$, prior work of Guth, Maldague, and Urschel (SIAM Matrix Analysis and Applications'25) can be combined with known sparsification techniques to give a $d^{1/6}$-approximation. We improve over these results by polynomial factors, giving a $d^{1/8}$-approximation. Moreover, we construct sum-of-squares certificates for the $2 \rightarrow q$ norm. This directly implies improved algorithms for robust mean and covariance estimation, robust regression, and clustering, when the data only satisfies a bound on its $q$-th moment. --- ### A Principled Self-Referenced Early Stopping Approach for Deep Image Prior Authors: Chaoyan Huang, Cheng-Han Huang, Ismail R. Alkhouri, Rongrong Wang Published: 2026-05-24 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2605.25299v2 Recently, Deep Image Prior (DIP) has demonstrated strong capabilities for solving inverse imaging problems (IIPs) by optimizing a randomly initialized convolutional neural network in a training-data-free regime. However, DIP suffers from overfitting to noisy measurements due to network over-parameterization, making early stopping (ES) essential. The most successful ES method tracks fluctuations in the running variance of the network output to detect overfitting. However, in many applications, these fluctuations may appear prematurely, leading to unstable reconstructions. In this paper, we first show that nearly optimal DIP early stopping can be achieved when two independent noisy copies of the degraded image are available. Motivated by this observation, and since obtaining two fully independent copies is infeasible, we propose an overfitting detection framework based on constructing pseudo self-referenced images, resulting in three IIP-specific algorithms. Our approach is further supported by theoretical results on single-reference validation, pseudo-validation estimation, and the impact of shared noise. Across different IIPs, ranging from natural image restoration to medical image reconstruction, and under varying noise levels and noise types, our methods consistently outperform existing DIP early stopping approaches, all without requiring an accurate estimate of the noise level. --- ### Eureka: Intelligent Feature Engineering for Enterprise AI Cloud Resource Demand Prediction Authors: Hangxuan Li, Renjun Jia, Xuezhang Wu, Yunjie Qian, Zeqi Zheng, Xianling Zhang Published: 2026-05-24 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.25297v2 Effective features are crucial for predictive model performance, but creating them often requires domain expertise, limiting scalability across applications. We define feature engineering as an agentic code generation problem: features are not static data transformations, but executable programs that can be generated, evaluated, and iteratively improved. We present Eureka, an LLM-driven framework with three stages. (1) An Expert Agent, fine-tuned via SFT on domain knowledge, produces structured feature design plans in JSON format. (2) An LLM Feature Factory translates each plan into executable Python code through chain-of-thought reasoning, turning feature hypotheses into runnable programs. (3) A Self-Evolving Alignment Engine uses Reinforcement Learning (GRPO) with dual-channel reward (metric-based utility + semantic alignment) to enhance code quality. By expressing features as programs, the learned generation patterns can transfer across domains. Evaluated on 7 public benchmarks in healthcare, finance, and social domains, Eureka consistently outperforms both traditional AutoFE and LLM-based baselines. We further demonstrate Eureka's effectiveness on cloud GPU resource demand prediction at Alibaba Cloud, where Eureka improves demand fulfillment rate by 16% and lowers computing resource migration rates by 33%. --- ### Geometry-Aware Image Flow Matching Authors: Junho Lee, Kwanseok Kim, Joonseok Lee Published: 2026-05-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.25294v1 Recent advances in generative models highlight the power of geometry-aware modeling in manifold-constrained settings. Yet, for natural images, the field remains confined to Euclidean assumptions, failing to exploit the potential of intrinsic geometric structures within the data. In this work, we investigate the geometry of natural images and observe that semantic information is predominantly encoded in directional components, while norm components can be approximated by the global average. This property holds across both RGB and latent spaces, suggesting that natural images can be effectively modeled on a hypersphere. Building on this finding, we introduce Spherical Optimal Transport Flow Matching (SOT-CFM), which utilizes angular distance, and Spherical Flow Matching (SFM), which constrains dynamics directly on the manifold. Our experiments demonstrate that these geometry-aware methods achieve superior performance against Euclidean baselines. Ultimately, this work provides a novel perspective that bridges the gap between Riemannian manifold-based modeling and natural image generation. --- ### Neuromorphic LiDAR-based Bird's Eye View Object Detection using Energy-efficient Spiking Neural Networks Authors: Sambit Mohapatra, Senthil Yogamani, Heinrich Gotzig, Patrick Mader Published: 2026-05-24 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2605.25293v1 Autonomous driving perception demands accurate and efficient processing of three-dimensional sensor data under strict power constraints. Traditional convolutional neural networks achieve strong detection accuracy but are computationally intensive, limiting their suitability for deployment on resource-constrained neuromorphic platforms. Spiking neural networks offer a compelling alternative through event-driven sparse computation, yet their application to complex real-world perception tasks such as three-dimensional object detection remains limited. In this work, we propose an end-to-end spiking encoder-decoder network for object detection in bird's eye view representations of LiDAR point clouds, trained using surrogate gradient backpropagation. We train two variants: a membrane potential variant that reads continuous neuron state at the output stage for maximum accuracy, achieving $92.05$/$87.04$/$86.51$ AP at $\mathrm{IoU}\!=\!0.5$ (Easy/Moderate/Hard), and, a fully binary spiking variant that operates exclusively on spike trains at every layer for direct neuromorphic deployment. We evaluate four input spike encoding strategies and demonstrate that allowing the network to learn spike representations directly from data outperforms hand-crafted Poisson, latency, and z-axis encoding schemes on the KITTI benchmark, where sequential frames are unavailable and the BEV input is presented repeatedly across timesteps as a proxy for temporal streaming. A block-wise energy analysis demonstrates a $3.33\times$ reduction in synaptic operation energy over an equivalent CNN under conservative loop-based operation. Together, these results demonstrate the viability of spiking neural networks for accurate and energy-efficient neuromorphic perception in autonomous driving. --- ### Choosing Online Experiment Designs under Interference in Ads, Recommendations, and Member-Experience Systems Authors: Prashant Shekhar, Caroline Howard Published: 2026-05-24 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2605.25290v1 Online experiments in ads, recommendation, and member-experience systems are often planned before the dominant interference mechanism is known. A treatment may propagate through budgets, inventory, producer exposure, graph spillovers, or temporal carryover, making the randomization design itself a statistical decision. We formulate this problem as robust design selection over uncertain exposure mechanisms. Given a finite catalog of six implementable designs, the selector compares each design by worst-case planning risk over an ambiguity set. The risk combines exposure bias, assignment-unit variance, minimum detectable effect, contamination or carryover, operational cost, and estimand mismatch. For theoretical justification, the paper develops a geometry-aware guarantee, stating that design bias is bounded by Wasserstein distance to the launch exposure distribution, and this penalty is minimax tight under Lipschitz exposure response. We also prove finite-catalog approximation and a robust selector theorem with excess-risk control, exact recovery under separation, and certified shortlists when the risk surface is flat. Empirically, the same selector gives different recommendations across samples from public datasets. It selects user-randomization on Criteo ads with dimensionless robust risk 1.295, switchbacks on Open Bandit-bts/men with risk 2.105, and cluster-randomization on KuaiRand with risk 2.240. The Open Bandit case stresses known but uneven logging support, with propensities from 0.00006 to 0.594 and a 5.17% IPS effective-sample share. Overall, the paper contributes an interference-aware experiment design framework based on mechanism-robust design decisions, where the output is either a justified design choice or an uncertainty shortlist. --- ### Knowing but Not Showing: LLMs Recognize Ambiguity but Rarely Ask Clarifying Questions Authors: Jinyan Su, Claire Cardie Published: 2026-05-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.25284v1 User queries are often underspecified and may admit multiple valid interpretations. Rather than silently making assumptions about the user's intent, a helpful assistant should surface such ambiguity by asking a clarifying question. Doing so requires two abilities: recognizing that a query is ambiguous, and acting on that recognition by seeking clarification instead of answering directly. To study these abilities, we evaluate models on ambiguous, unambiguous, and disambiguated questions in three settings: standard question answering, explicit ambiguity judgment, and behavioral analysis, where a judge model classifies responses as direct answers, refusals, or clarifying questions. We find a clear gap between recognition and behavior: models often identify ambiguity when explicitly asked to judge it, yet in the QA setting they overwhelmingly default to direct answers. Retrieved context further widens this gap by improving answerability while making models even less likely to ask clarifying questions. --- ### READER: Reasoning-Enhanced AI-Generated Text Detection Authors: Pingfan Su, Kai Ye, Shijin Gong, Erhan Xu, Jin Zhu, Giulia Livieri, Chengchun Shi Published: 2026-05-24 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.25281v2 Recent advances in large language models (LLMs) have made it increasingly difficult to distinguish human-written text from AI-generated content. Many existing detectors train supervised neural classifiers that achieve strong in-distribution performance but are often opaque and can degrade substantially under distribution shift. We present READER, a reasoning-enhanced AI text detector that outputs both a human/AI label and a structured rationale describing the evidence for its decision. A key component of our approach is READ, a curated supervision set of rationales and verdicts. We fine-tune an LLM on READ to build READER, which reasons before detecting at inference time. Despite having only 1.5B parameters, READER consistently outperforms existing detectors as well as prompted, high-capacity LLM baselines (GPT-5.2, Gemini-3-Pro, and DeepSeek-V3.2), which are 100 to 1000 times larger in scale. ---