arXiv:2606.30249v2 Announce Type: replace Abstract: Detecting communities in heterophilic graphs -- where connected nodes often belong to different classes -- is hard for unsupervised methods: classical modularity and spectral methods are feature agnostic, while deep graph-clustering methods rely on contrastive or generative machinery that is opaque. We propose Curvature-Guided Sheaf Diffusion (CGSD), a fully unsupervised community-detection algorithm that uses the discrete Forman--Ricci curvature of each edge as its single topological signal, propagated through every stage of an end-to-end pipeline. CGSD makes three concrete contributions: (i)~a curvature-gated sheaf-diffusion encoder that gates edge messages by $\sigma(\kappa_e)$ and is trained from three label-free structural losses (modularity, anti-collapse, curvature-weighted reconstruction); (ii)~a curvature-aware spectral clusterer (CSpec) that re-weights the $k$-NN affinity of the embedding by $\sigma(\alpha \kappa_{e^*})$ before Ng--Jordan--Weiss; and (iii)~a unified label-free evaluation against nine truly-unsupervised baselines. On five heterophilic benchmarks (Cora, Cornell, Texas, Wisconsin, Chameleon), CGSD wins outright on Wisconsin and Chameleon and is competitive on the remaining three against nine unsupervised baselines. The gain over the strongest baseline is driven by the clusterer, not the encoder: on the same embedding, CSpec improves mean NMI from $0.091$ with $K$-Means to $0.107$ ($+15\%$, paired $t$-test $p=0.008$). The mechanism is interpretable: intra-community and inter-community curvature distributions are visibly separated. Code is open-sourced at https://github.com/woodywff/cgsd.
Science Journals
arXiv:2607.04139v1 Announce Type: new Abstract: Self-supervised learning (SSL) shows strong potential for cross-dataset transfer by improving feature representation and generalization. However, its application to EEG-based emotion recognition remains largely unexplored. Existing SSL methods struggle to capture the intricate spatiotemporal dependencies of EEG signals under varying channel configurations, extract fine-grained representations resilient to noise, and derive global features that generalize well across subjects. To address these challenges, we propose Masked Generative-Contrastive Representation Learning (MGCRL), a novel SSL framework specifically designed for EEG-based emotion recognition. Built upon a region-aware spatiotemporal encoder, MGCRL integrates generative and contrastive learning to achieve both fine-grained and global discriminative representations for cross-dataset generalization. MGCRL introduces three key designs: 1) a spatiotemporal encoder that incorporates region-based graph convolution to capture localized spatial and functional relationships, enhancing region-specific feature learning and mitigating the impact of varying EEG channel configurations across datasets; 2) a generative learning mechanism based on the joint embedding predictive architecture (JEPA) that utilizes masked features to capture noise robustness fine-grained representations, improving the model's capability to characterize subtle emotional states; and 3) a contrastive learning strategy that leverages masked and original features to learn temporally stable and cross-subject-invariant representations across the same stimuli, boosting emotion discrimination and cross-subject generalization. Under these designs, MGCRL exhibits remarkable ability to learn universal representation. Extensive experiments involving pretraining on the large FACED dataset and fine-tuning on multiple SEED-series datasets demonstrate the effectiveness of MGCRL.
arXiv:2607.04220v1 Announce Type: new Abstract: Empirical research on social media and politics has primarily treated platforms as distributive systems that expose users to particular messages. The mediatization literature, however, suggests shifting attention upstream: from circulation to production. Under intense competition for platform attention, political actors who depend on visibility face pressure to learn from recurrent differences in reach and engagement - shaping politics around platform media logic. This paper examines that production-side dimension of platforms political impact by introducing attention price analysis: an exploratory method for estimating the differentiated attention returns associated with forms of expression. Using RoBERTa reward models trained on residualized engagement across X/Twitter, Bluesky, and Mastodon, the analysis compares how platform environments reward rhetorical, emotional, epistemic, and relational features of public communication. The attention signal differs sharply across platforms and engagement actions. X/Twitter sharing rewards antagonism while penalizing respect and nuance; Bluesky reposting favors neutral, lower-emotion language; and Mastodon boosts reward reasoning, nuance, compassion, and collective expression. Toxicity is rewarded across platforms, but in bounded and nonlinear ways. The findings suggest that moving from X/Twitter to less engagement-optimized alternatives such as Bluesky and Mastodon does not eliminate attention pressures, but it may reward less antagonistic and more deliberative forms of politics. The paper contributes a production-side approach to social media and politics by making one dimension of platform media logic empirically visible.
arXiv:2607.04305v1 Announce Type: new Abstract: We develop a unified, certified lower bound on the time-to-boundary margin M for transient stability of interconnected dissipative systems under slow parameter drift. The companion work establishes M as the first-passage time of the joint state-parameter motion to the synchronism boundary and proves M = CCT exactly on the one-machine-infinite-bus reduction, while leaving the multimachine certified margin open. Here a composite (mixed-region) Lyapunov function, formed by absorbing the restoring intra-group coupling into group energy functions and treating only the residual cross-cut coupling through the comparison principle, yields a positively invariant inner estimate of the region of attraction whenever an associated test matrix is a nonsingular M-matrix. The certified region breathes with the drift: its size is governed by a single critical synchronising stiffness k_c (lambda), and as k_c -> 0 at the boundary the region breathes shut and the certified margin M_low <= M_true vanishes. We give a nonlinear sector form of the construction, a domain-neutral resilience-fragility reading in which the coupling that certifies order is the one whose growth certifies collapse, and a constructive control corollary establishing a sharp dichotomy between damping injection and structural action. The mechanism is demonstrated identically on the WSCC nine-bus power system and on an inertial Kuramoto network, whose normalised breathing curves collapse, to leading order, onto a single profile. We present this collapse as numerical evidence for a conjectured universal form; a normal-form proof is identified as the precise open step.
arXiv:2607.04153v1 Announce Type: new Abstract: Vision-based deep reinforcement learning involves dealing with high-dimensional inputs of image information. It is crucial to abstract effective states from high-dimensional image inputs and limited samples for sample-efficient reinforcement learning. To address this challenge, inspired by fields such as natural language processing and computer vision, we propose a self-supervised task based on mask prediction as an auxiliary task for reinforcement learning. This non-reconstruction method uses the sequence information collected by the agent from the environment and the context information in the sequence to predict the masked information, thereby strengthening the agent's understanding of the task and learning effective representations. Combined with transformers, we find that the model reconstructs the masked input sequence in the latent space. By feeding the compressed representations learned by this method into reinforcement learning models, we observe an improvement in the sample efficiency of reinforcement learning. Moreover, the model outperforms state-of-the-art sample-efficient reinforcement learning methods on multiple continuous and discrete control benchmarks.
arXiv:2607.04576v1 Announce Type: new Abstract: LLM agents increasingly answer questions against knowledge bases they help maintain. A common intuition holds that progressive disclosure, a compact catalog plus a one-line summary per page so the agent loads only what it needs, should make this cheaper than consulting a large monolithic index. We test that on a real 709-page markdown wiki maintained by an LLM. We retrofit it for progressive disclosure and run a preregistered ablation in which four versions of the corpus differ only in how the agent reaches the content: page bodies are byte-identical across arms, frozen as immutable git tags, so any measured difference is due to access structure alone. We cross the arms with three access conditions (a protocol-constrained agent, a free self-routing agent, and a catalog-preload regime) and grade answers blind against verified gold references with a cross-family judge. A pilot upended the premise: a capable tool-using agent never loads the index, inferring a page's path from the question and reading it directly, so the specific saving the retrofit targets does not materialize. We therefore made answer quality primary and cost secondary. Quality is non-inferior (the retrieval arm matches the index baseline within the preregistered margin) while cost falls in every regime, from about a third for a self-routing agent to well over half under catalog-preload, all confidence intervals excluding zero. The saving comes not from avoiding the index load but from more targeted access: the retrieval arm cites fewer pages and takes fewer tool turns. The study doubles as a case study in evaluation validity, applying threat-to-validity discipline to the tooling that produced it.
arXiv:2607.05242v1 Announce Type: new Abstract: Personalized incentive allocation is vital for e-commerce, where uplift modeling is the standard for estimating Individual Treatment Effects (ITE). However, traditional models often fail in complex multi-seller environments with violations of the Stable Unit Treatment Value Assumption (SUTVA). We identify two critical challenges: Seller-level Cannibalization, where incentives shift expenditure between shops without growing the platform, and Incentive-level Cannibalization, where organic conversions or alternative rewards introduce significant noise into incrementality estimation. In this paper, we propose CanniUplift, a unified framework to mitigate these dual-source cannibalization effects. Specifically, we design Platform-level Global Alignment (PGA) to capture cross-shop substitution through global GMV consistency constraints. To tackle incentive-driven noise, we introduce Redemption-based Decomposition Denoising (RDD), which uses redemption behavior to decompose treated outcomes and reduce attribution noise within an entire-space framework. Furthermore, a Treat-Attention mechanism is designed to model intricate interactions between users' historical behaviors and current treatment options. Extensive experiments on both synthetic and large-scale industrial datasets demonstrate that CanniUplift significantly outperforms state-of-the-art baselines. Ablation studies confirm that the integration of PGA and RDD consistently improves wAUUC and wQINI. Successfully deployed online, our framework achieved a 4.08% relative increase in platform-wide incremental GMV (Delta GMV) over the production baseline and improved ROI in online A/B tests, proving effective in driving global platform growth.
arXiv:2607.04178v1 Announce Type: new Abstract: Decentralized Finance (DeFi) lending protocols currently rely on heuristic, utilization-based bonding curves that mandate severe over-collateralization, systematically excluding under-collateralized assets like corporate invoices. This paper introduces a mathematically optimal pricing mechanism for decentralized credit: the Reverse Kelly Automated Market Maker (rkAMM), the core engine of our proposed lending framework. By inverting the Kelly Criterion, traditionally used for optimal bet sizing, we construct a dynamic interest rate discovery protocol that explicitly prices individual loan risk. The rkAMM ingests real-time Probability of Default (PD) streams from an off-chain Explainable AI oracle and dynamically calculates the exact interest rate required to sustain target liquidity provider (LP) yields. We mathematically derive the Reverse Kelly pricing function ($r = \frac{y + PD}{1 - PD}$), proving its strictly convex superiority over Aave and Compound's static utilization curves in managing capital efficiency. Furthermore, we deploy the rkAMM architecture via Solidity smart contracts, optimizing for gas-efficient 1e18 (WAD) floating-point arithmetic. To ensure decentralized transparency, our simulation infrastructure leverages MLflow for tracking yield hyperparameters, Data Version Control (DVC) linked to DagsHub for versioning Real-World Asset (RWA) data arrays, and localized edge-inference via Ollama (Llama-3) and Hugging Face (FinBERT) for zero-cost predictive modeling. Monte Carlo simulations across 10,000 macroeconomic stress scenarios confirm that the rkAMM maintains protocol solvency and stabilizes LP yields at 12-15\% net of expected credit losses. This work provides the foundational financial engineering required to bridge the \$2 trillion global supply chain finance gap using permissionless blockchain infrastructure.
arXiv:2607.02768v1 Announce Type: cross Abstract: Pathologic complete response and tumor shrinkage measure whether breast cancer responds to neoadjuvant therapy, but not whether that response was structurally favorable, persistent, or hidden beneath volume loss. We built an outcome-blind longitudinal DCE-MRI manifold from I-SPY2 trajectories to test whether pretreatment imaging carries a structural response phenotype missed by conventional descriptors. The dominant axis of response geometry was not recoverable from the full clinical and genomic stack -- age, receptor subtype, MammaPrint, PAM50, treatment arm, and tumor burden -- but became strongly recoverable once baseline structural entropy was added. A constrained representation mapping recovered the same axes as unconstrained decomposition, establishing the structure as intrinsic rather than a post-hoc interpretation. The phenotype persisted through therapy, and as treatment proceeded the volumetric signal faded while entropy stayed separated -- a crossover from burden to structural persistence. Among complete responders, structurally disordered tumors could shrink more early yet remain structurally disordered, a volumetric deception invisible to endpoint labels. External analyses in UCSF, I-SPY1, and Duke established recurrence relevance under representation-dependent boundaries, and a representation-family commensurability assessment showed why feature-name matching is insufficient: the same label can fail, transport, or entangle with extraction geometry. Pretreatment MRI therefore exposes a structural response phenotype that endpoint-based language leaves invisible -- including, among complete responders, a pretreatment imaging signal of structurally distinct response states that awaits prospective validation.
arXiv:2204.02803v2 Announce Type: replace Abstract: Sign language recognition from monocular video or 2D pose sequences is challenging, both because 3D information must be inferred from 2D observations and because the signal is inherently spatiotemporal. Moreover, the large and continually growing vocabulary of signs in production settings makes conventional closed-set classification impractical: adding a class requires new labeled data and retraining. We propose a contrastive Transformer-based model that learns rich representations of body key-point sequences, enabling direct comparison between embedding vectors. These representations support one-shot and few-shot tasks such as classification of signs never seen during training. On the LSA64 dataset, using only 48 classes for representation learning, the model reaches 88.4% accuracy on 16 held-out classes with as few as eight reference examples per class, and its accuracy improves consistently with the number of training classes and support examples.
arXiv:2606.30545v2 Announce Type: replace Abstract: 3D Gaussian Splatting (3DGS) has achieved remarkable success in real-time novel view synthesis, yet it suffers from severe overfitting under sparse-view settings due to insufficient geometric constraints. While recent methods introduce monocular depth priors to mitigate this, they inherently struggle with scale ambiguity and cross-view inconsistency, leading to defective geometry. In this paper, we propose StereoGS, a novel sparse-view 3DGS framework that integrates stereo priors to establish reliable binocular consistency. Unlike scale-agnostic monocular constraints, StereoGS introduces a Stereo Depth Regularization by constructing virtual stereo pairs during optimization and leveraging a foundation stereo model to enforce absolute scale and binocular-consistent structures. To further suppress overfitting and eliminate redundant primitives, we design a Gradient-Aware Opacity Decay strategy that dynamically penalizes Gaussians based on their relative opacity gradient magnitudes. Combined with a Consistency-Aware Dense Initialization using zero-shot multi-view depth estimation, StereoGS effectively anchors primitives to accurate scene surfaces. Extensive experiments on LLFF, DTU, Mip-NeRF360, and Blender datasets demonstrate that StereoGS achieves state-of-the-art performance in sparse-view settings without incurring any additional inference overhead. Project Page: https://stringerywh00.github.io/StereoGS_project_page/
arXiv:2606.30583v2 Announce Type: replace Abstract: Using 380 trillion tokens of realized AI consumption across more than four hundred large language models from the licensed proprietary OpenRouter dataset covering approximately 2 percent of current global monthly AI token consumption, we analyze how AI affects firms, markets, and workers. Leveraging the unprecedented size, scope and granularity data, we construct the AI Factor from growth in tokens, dollars, and users, estimate firm-level AI Betas from stock return comovement, and characterize the \emph{AI Premium}. First, we build a high-frequency AI factor and decompose it into salient components. Second, we show that firms whose returns covary more positively with the AI factor -- high AI beta firms -- earn higher subsequent returns, and the AI premium is large and heterogeneous. A value-weighted long-short strategy earns 64.1 basis points per week, and the premium is large for loadings on the intensive, frontier-oriented margin of AI consumption -- closed-source models, paying and seasoned users, and long prompts -- but not on casual or open-weight use. Third, the premium reaches beyond technology firms into consumer-facing and capital-heavy parts of the economy, but is absent in emerging markets, including China. Fourth, the AI exposure is more positive in nonroutine interactive work and more negative in analytical, scientific, and operations-control skills -- an occupation one standard deviation higher in interaction-and-communication content has 0.36-standard-deviation higher market-implied AI exposure. Additionally, we provide early evidence of the rise of the agentic economy.
arXiv:2606.30820v2 Announce Type: replace Abstract: Time Window Temporal Logic (TWTL) is a rich specification language for cyber-physical systems that can compactly express sequential tasks with explicit timing constraints. In this paper, we consider the problem of synthesizing control inputs for discrete-time linear systems subject to TWTL task specifications. Building on the quantitative semantics (robustness) recently introduced for TWTL in [1], we encode the robust satisfaction of a TWTL formula as a set of Mixed-Integer Linear constraints and pose synthesis as a Mixed Integer Linear Program (MILP) that maximizes the robustness degree. We prove that any feasible solution with positive objective value guarantees Boolean satisfaction of the specification. We address two synthesis settings: an \emph{open-loop} formulation that optimizes the full control sequence from the initial state, and a \emph{closed-loop} receding-horizon Model Predictive Controller (MPC) formulation that re-solves the MILP at each step using the current measured state. A key feature of our MPC formulation is a \emph{task-adaptive horizon} that exploits the TWTL Deterministic Finite Automaton (DFA) to determine the active sub-task at each step, limiting the prediction horizon to the remaining window of the current task rather than the full formula horizon, this makes each re-solve significantly cheaper than the initial open-loop solve.
arXiv:2606.30907v2 Announce Type: replace Abstract: In this experimental and theoretical study, we assess the extensional relaxation time of sodium alginate solutions by using dripping-onto-droplet capillary breakup rheometry (DoD), e.g., the capillary thinning and breakup of viscoelastic filaments formed following the coalescence of a millimetric-nozzle-generated pendant drop with a lower droplet cap of the same fluid contained in a millimetric pool in ambient air. Hence, we extend the analyses conducted by El Khoury et al. (2026) from Newtonian to viscoelastic fluids. Our approach relies on experiments recorded with a high-speed camera using sodium alginate in deionised water, with alginate concentrations ranging from 0.1% to 9% by weight. The results are depicted by considering the dynamics of fluid filament thinning, stress balances, and scaling laws. Extensional relaxation times are resolved from the filament diameter evolution. Three flow regimes are highlighted: capillary-inertial, capillary-elastic, and mixed capillary-inertio-elastic. The findings are summarised in a two-dimensional diagram that correlates the filament breakup time with different flow regimes using the important dimensionless parameter of the problem, e.g., the intrinsic Deborah number (which relates the extensional relaxation time to the characteristic capillary-inertial time). This diagram can be used to quantify both the solution's extensional relaxation time and the liquid/air surface tension solely from filament breakup times.
arXiv:2606.31399v2 Announce Type: replace Abstract: Water looks unchanged as it warms, then at a critical point it boils. We ask whether long-horizon language agents show an analogous transition in their implicit world models. In some parameter settings, changing state load by a small amount, or adding a single step of horizon, leaves behavior nearly unchanged; near a critical boundary, the same small change causes a sudden world collapse. We study this effect in a deterministic task family with exact per-step gold state. A large grid search over state cardinality, dependency density, horizon, branching, observation mode, and mutation rate reveals a phase diagram: a solved plateau, a narrow transition band, and a collapse floor. Per-step traces show the mechanism: world-state fidelity fails before action validity, so the agent is not merely choosing a bad action; it is acting from a corrupted world. Stronger models translate the critical boundary but do not remove the qualitative transition. These results make world-model collapse a measurable bottleneck for long-horizon agents.
arXiv:2606.31422v2 Announce Type: replace Abstract: Language agents acting over long horizons must maintain beliefs about tool states, object locations, graph edges, and subgoal dependencies. When these beliefs drift, failures can be fixed neither by longer reasoning traces nor by ordinary self-reflection, since the missing evidence lies in the environment. We formulate environment probing as a budgeted decision problem for structured agent world models: before acting, the agent may query the current value of one belief field, update its table, and pay one interaction step. We introduce EnvProbe, a simple scoring policy that combines task criticality, staleness, verbalized uncertainty, and dependency role. A type-stratified analysis separates the benefit of belief repair from the cost of displaced task actions and predicts different behavior for procedural and spatial beliefs. In three controlled environments with gold belief states, EnvProbe improves terminal world-state accuracy over periodic probing by 11.76 percentage points on procedural tool-dependency tasks, 3.79 points on spatial tasks, and 6.45 points overall. Ablations show that task-structural terms are the main source of the gains, while self-reported uncertainty is unreliable under confident wrong beliefs. The results suggest that agent calibration should be treated as an action-selection problem over environment evidence, not only as a model-internal reasoning problem.
arXiv:2606.31557v2 Announce Type: replace Abstract: In the evolving threat landscape, adversaries exploit software vulnerabilities to launch sophisticated attacks, challenging traditional defenses. Although databases like CVE and NVD provide detailed technical information, they often lack links to attacker behaviors such as tactics and techniques, limiting effective threat interpretation and response. This work bridges this gap by connecting vulnerabilities with behavioral patterns from the MITRE ATT&CK framework. We construct a CVE-TTP Knowledge Graph that links CVEs to tactics and techniques using classification and relation extraction. Transformer-based models are developed for behavior identification, with CySecBERT achieving macro F1-scores of 87.71% (techniques) and 96.16% (tactics). Also, we created an annotated dataset with 24,820 entities and 43,608 relations for entity and relation extraction. The pipeline-based approach achieves macro F1-scores of 0.86 (entity extraction) and 0.99 (relation extraction), while a span-based joint model achieves 0.78. These outputs are integrated into a Neo4j-based Cyber Threat Knowledge Graph, enabling structured visualization of vulnerabilities.
arXiv:2606.31672v3 Announce Type: replace Abstract: Despite rapid progress in interactive world models (IWMs), existing benchmarks evaluate action following only at trajectory level and ignore memory and interaction physics. We introduce WorldRoamBench, an open-world benchmark for long-horizon stability across four dimensions, each with tailored innovations: (i) Action: per-frame action metric bypassing cross-model semantic scale disparity and exposing failures hidden by trajectory; (ii) Vision: segment-based drift metric capturing non-monotonic mid-sequence collapse missed by start-vs-end comparisons; (iii) Physics: controllability-gated evaluation over mechanics, optics, and 3D consistency, scoring plausibility under faithful action execution; (iv) Memory: action-decoupled protocol evaluating scene memory via transition-localized 3D point-cloud reconstruction and subject memory via tracking-plus-VLM reasoning. The benchmark comprises 600+ test cases across Nature, Urban, and Indoor scenes in first/third-person views with WASD 10-60s continuous interaction. Evaluating 10+ open/closed-source models reveals none reliably satisfies all dimensions; even the best achieves only moderate scores. Advances on WorldRoamBench are steps toward IWMs that are stable, physically grounded, memory-faithful, and deployable in real-world applications.
arXiv:2606.31876v2 Announce Type: replace Abstract: To improve safety in Large Language Models (LLMs) we can either perform post-training alignment or exploit refusal directions in the activation space. Both strategies are less feasible in Multimodal LLMs (MLLMs) as they require unsafe multimodal data, harder to collect than their unimodal counterpart. In this work, we relax this constraint and investigate whether textual refusal directions, extracted directly from the LLM backbone, generalize across modalities (i.e., image, video). Preliminary findings confirm this ability, though effectiveness is conditioned by layer selection, steering strength, and cross-modal alignment, with the latter causing safe multimodal inputs to be spuriously steered toward refusal. Building on this, we introduce Modality-Agnostic Refusal Steering (MARS), a light-weight training-free approach that injects multimodal safety without the need for multimodal safety data. MARS corrects modality misalignment via activation re-centering, adaptively scales steering strength within a geometrically defined trust region, and selects the optimal intervention layer, operating at the first generated token. Evaluated on five SOTA MLLMs across safety, utility, and video jailbreak benchmarks, MARS achieves consistent safety gains while preserving utility. These results reveal that safety-relevant structure is shared across modalities and that textual refusal directions are a powerful and underexplored foundation for multimodal alignment.
arXiv:2606.31984v3 Announce Type: replace Abstract: Industrial recommendation systems serve billions of users through a multi-stage funnel -- retrieval, early-stage ranking, and re-ranking -- where the final re-ranking step disproportionately shapes user engagement and downstream performance, particularly for carousel and grid display formats. Despite growing enthusiasm for Large Language Models (LLMs) in recommendation, three gaps hinder industrial adoption: (1) most efforts target retrieval and ranking, leaving re-ranking -- the stage closest to the final user experience -- largely underexplored; (2) LLMs are typically deployed zero-shot or via supervised fine-tuning, underutilizing the reasoning capabilities unlocked by reinforcement learning (RL) on verifiable rewards; (3) deployed catalogs index billions of items with non-semantic identifiers that lie outside any base-LLM vocabulary. We present GR2 (Generative Reasoning Re-Ranker), an end-to-end framework that combines (i) mid-training on semantic IDs produced by a tokenizer with >=99% uniqueness, (ii) reasoning-trace distilled from a stronger teacher via targeted prompting and rejection sampling, and (iii) RL with verifiable rewards purpose-built for re-ranking. To make GR2 resource-viable, we further (iv) introduce a context compressor that amortizes training cost, On-Policy Distillation (OPD) as a scalable alternative to SFT -- which we find collapses at industrial scale -- and reasoning distillation for low-latency serving. GR2 delivers +18.7% R@1, +7.1% R@3, and +9.6% N@3 over legacy baselines on industrial-scale traffic. We further find that reward design is critical in re-ranking: LLMs often hack rewards by preserving the incoming order or exploiting position bias, motivating conditional verifiable rewards as essential industrial components.
arXiv:2606.31986v2 Announce Type: replace Abstract: Chain-of-thought (CoT) reasoning has enabled multi-modal large language models (MLLMs) to tackle complex visual reasoning tasks by generating explicit intermediate reasoning steps in natural language. However, this text-based reasoning paradigm is inherently slow at inference time with even thousands of tokens and fundamentally constrained by the expressiveness of natural language. In this paper, we propose CoLT, (Chain of Latent Thoughts), a novel framework that teaches multi-modal models to reason through a chain of latent thought representations instead of verbose text tokens, which can perform thinking with as few as 3 steps. Naively forcing the model to think with latent states easily produces meaningless semantics and makes training unstable. To effectively regulate the latent reasoning process, we introduce a lightweight external decoder that provides step-level supervision for each latent reasoning step in two complementary directions: a forward mode that decodes latent thoughts into the textual reasoning of the next step, and a backward mode that aligns decoder hidden states with the model's latent thoughts given preceding textual context. We further incorporate internal supervision that encourages coherent step-by-step latent transitions. The decoder and internal supervision are removed during inference to maintain high efficiency of latent reasoning. Extensive experiments on eight benchmarks demonstrate that CoLT not only outperforms existing latent reasoning methods such as CODI and SIM-CoT, but also surpasses latent visual reasoning approaches that rely on auxiliary images with costly annotation requirements. Compared to text CoT methods, CoLT can notably reduce the inference time by 10.1$\times$ and text decoding time by 22.6$\times$. Code is released at https://github.com/hulianyuyy/CoLT.
arXiv:2607.04277v1 Announce Type: new Abstract: The pursuit of self-evolving AI raises a critical question: when is autonomous self-improvement sustainable rather than degenerative? Drawing an analogy to von Neumann's complexity threshold for self-reproducing automata, we argue that sustainable recursive self-improvement in Large Language Models (LLMs) requires a functional analogue: introspection -- the system's capacity to simulate its own operations and target modifications. Grounded in Kleene's Second Recursion Theorem, we demonstrate the theoretical existence of such introspective programs. However, an empirical review reveals that while current LLMs exhibit quasi-introspection (e.g., partial metacognition), they fall short of true introspection due to structural bottlenecks: a lack of complete self-access, the feedforward nature of the Transformer, and computational class constraints that prevent fixed-point iteration. We conclude by outlining architectural paths to cross this complexity threshold and discussing the associated safety implications.
arXiv:2603.05363v3 Announce Type: replace Abstract: In realistic pursuit-evasion scenarios, abrupt target maneuvers generate unavoidable periods of elevated uncertainty that result in estimation delays. Such delays can degrade interception performance to the point of causing a miss. Existing delayed-information guidance laws fail to provide a complete remedy, as they typically assume constant and known delays. Moreover, in practice they are fed by filtered estimates, contrary to these laws' foundational assumptions. We present an overarching strategy for tracking and interception that explicitly accounts for time-varying estimation delays. We first devise a guidance law that incorporates two time-varying delays, thereby generalizing prior deterministic formulations. This law is driven by a particle-based fixed-lag smoother that provides it with appropriately delayed state estimates. Furthermore, using semi-Markov modeling of the target's maneuvers, the delays are estimated in real-time, enabling adaptive adjustment of the guidance inputs during engagement. The resulting framework consistently conjoins estimation, delay modeling, and guidance. Its effectiveness and superior robustness over existing delayed-information guidance laws are demonstrated via an extensive Monte Carlo study.
arXiv:2607.04290v1 Announce Type: new Abstract: Large Language Models (LLMs) are proposed as control interfaces for next-generation networks, but their latency, hallucinations, and lack of control guarantees make them unsuitable for near-real-time packet schedulers, especially in dynamic V2X environments. This paper introduces Agentic-V2X, an architecture where a small, locally deployed language model acts as a periodic non-real-time rApp-inspired policy creator, while a lightweight xApp-like controller executes validated policies at intervals suitable for scheduling. The framework targets deadline-aware 5G NR V2X scheduling with heterogeneous services (teleoperated driving, cooperative awareness, HD map sharing, and sensor sharing). Given a scenario summary, service objective, and telemetry, the LLM generates a structured policy containing service priorities, weight bounds, and safety constraints. A validator checks and repairs the policy before the controller enforces it via scheduler-weight adaptation in ns-3/ns3-ai. The evaluation compares proportional fair scheduling, static expert policies, a heuristic xApp, static LLM policies, and adaptive LLM-rApp policies over 126 completed runs. Metrics include deadline-constrained packet reception ratio, tail latency, deadline violations, throughput, fairness, policy validity, and safety interventions. Results show that the adaptive LLM-rApp/xApp design generates valid and executable policies and remains competitive at several operating points, including improved mean critical reliability over PF at the highest density. However, paired statistical analysis shows that the adaptive method is not the best aggregate method and remains below the strongest static policies overall. These results support Agentic-V2X as a safe, executable small-LLM policy-generation architecture rather than a universally dominant scheduler.
arXiv:2607.04591v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models have demonstrated strong capabilities in robotic manipulation by integrating visual perception, language understanding, and robot action generation. Existing research has primarily focused on improving model architectures, training strategies, and dataset scale, while little attention has been paid to how demonstrations are collected and organized. We identify demonstration organization as a fundamental yet overlooked aspect of imitation learning, as it directly affects policy learning efficiency, training stability, and policy generalization. To address this gap, we propose a simple-to-complex structured demonstration collection strategy for VLA learning using a dual-arm robotic platform. Our approach systematically organizes data through three general principles: (i) decomposing complex manipulation tasks into progressively learnable sub-skills, (ii) standardizing the interaction environment to reduce unnecessary variability, and (iii) organizing demonstrations according to progressively increasing task complexity. This structured design enables VLA models to first acquire fundamental manipulation skills before learning increasingly complex task compositions, facilitating more effective learning of long-horizon manipulation tasks. We evaluate the proposed strategy on two representative robotic manipulation tasks: block grasping and sorting, and towel folding. Experimental results show consistent improvements in task success rate and training stability compared with the baseline method of directly collecting end-to-end complete task trajectories. These findings highlight demonstration organization as a previously underexplored but important factor in VLA learning and provide practical insights into efficient skill acquisition, scalable dataset construction, and long-horizon robotic manipulation.