Papers

Existing model marketplaces typically trade and select individual models as indivisible units, limiting their ability to exploit complementarities among heterogeneous experts. This paper proposes an MoE-based model market framework that lifts Mixture-of-Experts from a model-level learning architecture to a market-level coordination mechanism. In this framework, brokers use gating networks to coordinate multiple heterogeneous experts and deliver a composite model service. We formalize the market participants, service workflow, expert cost structure, and a welfare objective that combines predictive utility with heterogeneous execution costs. We then derive a cost-aware gating mechanism and market-aware training objective, and introduce a cost-adjusted revenue allocation rule that distributes residual revenue according to realized expert participation and execution cost. We also establish basic theoretical properties of the allocation rule, including budget balance, participation monotonicity, and cost sensitivity. Experiments over five random seeds on fifteen tabular and image benchmarks use independently trained and frozen neural and tree-based experts together with latency-derived execution costs. MoE Market achieves the highest mean welfare on all fifteen datasets and a lower mean expected cost than Standard MoE in every case, while maintaining competitive predictive performance. The allocation experiments further demonstrate systematic sensitivity to expert participation and cost, together with substantially lower computational overhead than exact Shapley allocation. These results suggest that MoE can serve as a market-level coordination principle for collaborative, cost-aware, and economically grounded model marketplaces.

Hallucination remains a significant challenge in Large Vision-Language Models (LVLMs). Existing training-free methods generally mitigate hallucinations through contrastive decoding or visual enhancement, often increasing the relative influence of visual evidence during generation. This raises a fundamental question: Can LVLMs dynamically regulate the contributions of different context sources to suppress hallucinations? In this work, we investigate and quantify how LVLMs coordinate multiple context sources during decoding and examine how this intrinsic behavior can guide hallucination mitigation. We find that LVLMs exhibit an intrinsic vision-attending tendency that can guide adaptive visual steering, while textual contexts can also contribute to hallucination mitigation. Motivated by these findings, we propose AIMS (Adaptive Information Multi-source Steering), a lightweight training-free framework that adaptively coordinates visual, prefilled textual, and generated contexts during decoding. Specifically, AIMS constructs compact prototypes for the three context domains and estimates their affinities with the current query to determine head-wise steering weights. The resulting multi-source steering direction is applied to the query representation, enabling adaptive context integration without additional model training or auxiliary forward passes. Extensive experiments across multiple LVLMs and decoding strategies demonstrate that AIMS effectively mitigates object hallucination while maintaining competitive general-purpose multimodal capabilities.

We consider the problem of learning linear dynamical systems under adversarial contamination from a single trajectory of length $T$. While identification of linear dynamical systems itself is well-studied, the problem of robust system identification under adversarial contamination is relatively less explored. In this work, we study the setting where a fraction of the $T$ observations are contaminated by adversarial outliers. We propose different estimators based on relaxations of least-trimmed squares along with an alternating minimization algorithm. Furthermore, we also propose two estimators which exploit the group-sparsity (through penalization/hard-constraints) of the outliers. For the estimator with group-sparse penalty, we derive non-asymptotic error bounds which establish its robustness to outliers. We also show empirically that the proposed estimators work well in practice.

Combining a reconfigurable intelligent surface (RIS) with a receive antenna array is a promising low-complexity architecture for multi-user uplink reception, but its performance analysis for more than two users has remained an open problem: the zero-forcing (ZF) signal-to-interference-plus-noise ratio (SINR) no longer admits an explicit, low-dimensional closed-form expression, and its distribution is analytically intractable for design purposes. This paper addresses this gap for a K-user, K-antenna uplink system in which a large-scale, L-element RIS, partitioned into K user-dedicated sub-surfaces, precedes ZF reception at the base station. Through an asymptotic analysis in which the sub-surface sizes grow without bound, we show that the orthogonal projector underlying the ZF SINR converges to a rank-one matrix aligned with the desired user's channel. We use this convergence to derive a closed-form asymptotic approximation for the average per-user SINR that depends only on deterministic channel parameters. Treating this expression as a tractable design objective, we prove that equal partitioning is approximately sum-rate-optimal at leading order regardless of path-loss asymmetry across users. We also develop a low-complexity greedy pairwise-transfer search that refines the partition beyond this leading-order optimum. Monte Carlo simulations across a range of system and RIS sizes confirm that the closed-form SINR tracks the exact simulated rate closely once the RIS is large relative to the number of users. The greedy search also yields consistent, if modest, sum-rate gains over equal partitioning, validating the theory as both an accurate performance predictor and a practical design tool.

Stance detection requires identifying an author's attitude toward a given target, sometimes based on conversational context. Jev, a specialized decision model designed for structured decision-making, offers an alternative to general-purpose large language models (LLMs). In this work, we evaluate Jev on two stance detection datasets, VAST (English texts) and ZS-CSD (Chinese conversations), comparing it with four general-purpose LLMs and two fine-tuned models. Results show that Jev achieves competitive performance on VAST, matching GPT-5.6 and outperforming the other general-purpose LLMs. However, it falls behind stronger LLMs on ZS-CSD, particularly in distinguishing favor from against. Further analysis suggests that this limitation may be related to understanding reply relationships and stance direction rather than conversation length alone. These findings highlight both the potential and limitations of Jev for stance detection.

This paper considers the reach-stabilize prob- lem for a class of nonlinear control-affine systems with unknown parametric uncertainties, where the system states must remain in a safe set at all times, or enter a safe set in finite time and then remain in it, and converge to a goal point. An estimator is designed that generates a parameter estimate and a computable, nonincreasing bound on the estimation error from a known initial error bound, without requiring persistence of excitation. The bound defines margins that are added to a control barrier and a control Lyapunov function condition, which are then enforced in a quadratic program for efficient control design. It is shown that, for the closed-loop system, the safe set is forward invariant when the system starts within the safe set and is finite-time reachable from outside the safe set with a recovery-time bound that does not depend on the unknown parameter, and that the goal point is exponentially stable. In two numerical examples, with the initial state outside and inside the safe set, the proposed control policy, a baseline oracle control policy that uses the true parameter, both recover or remain in the safe set and converge to the goal point, while another baseline control policy that uses a fixed, incorrect, parameter does not remain

Large language models (LLMs) are increasingly embedded as components in software systems, marketed under labels such as chatbot, copilot, retrieval-augmented generation, workflow, coding agent and AI agent. Whether these labels denote genuine architectural forms or serve as branding has not been assessed systematically. In the sources surveyed, labels do carry architectural content, most clearly in vendor usage: copilot denotes a router-worker architecture operating a host application under step-by-step user confirmation, while the more recent shift to the label agent coincides with AI-planned multi-step execution of which the user sees only the outcome. The coding agents of four major providers share one architecture, a reason-and-act loop delegating to subagents. This survey describes seven recurring forms—LLM chats, custom agents, retrieval-augmented generation (RAG), AI-enhanced workflows, copilots, coding agents, and, in part, agentic RAG—in a common vocabulary of agents and tools. Each is characterized along four structural dimensions (agentic RAG only partially): the architectural pattern, the control of execution and the point of user intervention, the number of agent calls per task, and tool use. An illustrative corpus of 22 systems from research publications and vendor documentation grounds the descriptions and shows where they reach their limit.

Assembly documentation is a downstream manufacturing artifact that is still usually authored by interpreting CAD models by hand. Structured product data and large language models are both available, yet studies of CAD interpretation, assembly sequence planning, instruction writing, and human oversight have largely proceeded separately. This paper formulates CAD-grounded assembly instruction generation: the production of natural-language assembly procedures constrained by structured engineering information extracted from CAD models. The proposed framework maps a STEP assembly to a typed ProductGraph intermediate representation, derives a precedence order by deterministic topological sorting, realizes each step as language conditioned only on selected graph context, attaches per-step visual documentation, and applies rule-based and model-assisted checks. PDF export remains disabled until a human reviewer resolves every quality flag. The case study establishes endto-end feasibility on a built-in six-part reference assembly: the pipeline preserves a reported assembly order and carries quantity, material, and torque into an exported manual page. Generalization and geometric validation remain open empirical questions. The contribution is an architecture that separates engineering state, deterministic reasoning, grounded language realization, verification, and human release.

The analysis of spatiotemporal event data is essential for informed decision-making in domains such as disaster response, conflict analysis, or intelligence investigations. However, the complexity and interdependence of spatial, temporal, and multiple thematic attributes pose significant challenges for both analysis and visualization. While space-time cubes (STCs) present a powerful integrated visualization technique to analyze this kind of data, existing approaches often lack support for complex exploratory workflows, thus limiting the ability to derive meaningful insights. We address this gap by introducing STcubeOperator, a novel framework that models analysis tasks through space-time cube operations, considering them in context of visualizations, interactions, and computational choices, and implement them in an interactive visual analytics environment. By expressing analysis tasks as a sequence of multiple elementary operations–such as filtering, chopping, and flattening–our approach enables analysts to dynamically explore data from different perspectives. We further provide an open-source prototype implementing the operations in a 3D interactive environment to facilitate task-based exploratory analysis of spatiotemporal event data. We demonstrate the applicability of our framework with a case study based on real-world data on strategic and military operations in the Russia-Ukrainian War, showing its capabilities to reveal spatiotemporal patterns. An expert user study (n=8) shows how specific tasks can be solved with our framework, highlights the versatility of our approach, and provides valuable insights on which operations experienced analysts utilize in practice.

Retrieval-Augmented Generation (RAG) systems extend large language models (LLMs) with external knowledge through a multi-stage pipeline. While this architecture can improve the factual grounding of generated answers, it introduces structural attack surfaces that extend beyond those of standalone LLMs. In this paper, we introduce a security meta-model that captures explicit causal relationships between RAG surfaces, attacks, weaknesses, risks, and CIA impact (Confidentiality, Integrity, Availability). Its purpose is to provide security engineers with a structured and user-friendly framework for gathering and assessing the risks, weaknesses, and mitigations relevant to their RAG deployment. We designed the meta-model through an iterative, structured analysis of 43~publications (2023–2026) and instantiated it as a catalog populated with the security threats and remediations reported in the literature. Filtering the catalog according to a deployment configuration produces a risk profile containing the risks applicable to that deployment. An interactive web visualizer lets users navigate the catalog as a graph, follow causal chains, and explore stakeholder-specific views. Analysis of the catalog revealed a persistent imbalance between attack-focused and defense-focused research, a concentration of threats at ingestion, and coverage gaps affecting output integrity. Coverage is assessed against the OWASP LLM Top~10, and operational applicability is illustrated across textual, graph-based, and multimodal RAG configurations.

We present a benchmark for predicting final output and checkpoint state from source and input alone. It extends CRUXEval-style output prediction with paired shorter- and longer-trace inputs and checkpoints inside and after a loop. The benchmark contains 400 cases from 371 Python and C++ programs, evaluated under seven settings from four model families without tools or code execution. Of 11,200 planned predictions, 11,151 produced gradable responses. Reasoning-enabled settings outperform their off counterparts by 33.1 to 55.2 percentage points on completed responses. The strongest setting scores 93.0% on shorter-trace final output, 77.0% on longer-trace final output, and 65.5% and 63.5% on the two state tasks; these scores also hold when missing responses count as wrong. Across 2,397 matched Python comparisons with identical source, changing to the longer-trace input yields 528 correct-to-wrong changes and 147 reversals. Source-clustered analyses preserve this accuracy gap, while adjusted Python models give no evidence of a positive incremental association between cumulative state load and error. Changed inputs and checkpoint tasks alter several factors together, so the gaps do not isolate trace length or an internal state-tracking mechanism. The benchmark exposes errors hidden by short-output scores alone.

We construct a family of scalar-closed, non-additive maximum sum-rank distance codes (MSRD) by switching norm components of a $\sigma$-rational normal curve and lifting the resulting point set through a linearized Reed–Solomon syndrome map. We characterize the admissible switches by a multiplicative stability condition on the norm classes. In particular, for $1 \leq n_i \leq m$, $N=n_1+\ldots+n_{\ell}$ and $2\leq\delta\leq N-1$, every partition of $\mathbb{F}_q^*$ satisfying a specific condition, referred to as Condition $(\diamond)$, yields a code in $\mathbb{F}_{q^m}^{N}$ of size $q^{m(N-\delta+1)}$ and minimum sum-rank distance $\delta$; the code is non-additive whenever a curve component is retained. For $\ell\geq2$ these appear to be the first non-additive MSRD codes in the literature, all previously known families with more than one block being $\mathbb{F}_q$-linear. Moreover, every code of the family has the same sum-rank weight distribution as an $\mathbb{F}_{q^m}$-linear MSRD code with the same parameters, although the geometry of the switched set distinguishes it from linearized Reed–Solomon codes. In the single-block case, an explicit rank isometry identifies the construction with the cone codes of Durante, Grimaldi and Longobardi.

The definition and quantification of fuel (or energy) poverty remain contentious, particularly within the English policy landscape. The current official definition - Low Income Low Energy Efficiency (LILEE) - systematically underestimates the condition in low- and middle-income households, effectively obfuscating changing energy affordability norms. This paper presents a sensitivity analysis of fuel poverty trajectories in Nottingham, UK, across competing definitions and diverging optimistic domestic efficiency upgrades) and pessimistic (energy price inflation) scenarios. Results demonstrate that LILEE overstates the efficacy of efficiency upgrades while failing to adequately reflect the impact of price shocks. To address these deficiencies, we propose the Temporal Equity Framework (TEF), a budget standard-based approach paired with a proportional energy expenditure indicator that captures shifting affordability norms more equitably. The TEF also introduces continuous expenditure-based depth indicators to evaluate the entrenchment of fuel-poor households (the fuel poverty gap) and the resilience of non-fuel-poor households (the fuel poverty buffer). We argue that responsive definitions which also recognise fuel poverty as a continuum of vulnerability, rather than a static, dichotomous state, are paramount to ensuring the equitability of energy transitions. Further, we suggest that the early identification of fuel poverty and spatial targeting of remedial resources could be markedly improved by adopting the TEF. We conclude that the adoption of a multidimensional budget standard-based definition is essential to facilitate a more accurate and just evaluation of energy transitions in England.

Music understanding often requires comparing excerpts and reasoning about relationships among songs, sections, and stems. However, existing large audio-language models (LALMs) and music question-answering datasets typically operate on single recordings or compare independently sampled tracks with no known production relationship. We introduce STEMMA, a multi-audio music question-answering framework built around production provenance: whether excerpts originate from the same track or section, and which stems belong to which mixtures. Because such relations are sparse under conventional audio-first sampling, STEMMA adopts a relation-first construction strategy: it first specifies a target relation and then queries the catalog for excerpts that satisfy it and hard negatives that do not. Labels are determined directly from catalog provenance rather than generated by a language model from metadata. We build STEMMA-Bench for evaluation and a track-disjoint training set, STEMMA-Instruct. Fine-tuning two LALMs on STEMMA-Instruct improves multi-audio reasoning, with the largest gains on structural relations directly determined by the catalog, while preserving single-audio music understanding.

Automated research systems support literature synthesis, ideation, experiments, writing, and peer review, but their evaluation is dispersed across tasks, benchmarks, and studies that are difficult to compare directly. We review this literature from the perspective of evaluation design and evidence, covering six targets: literature synthesis, research ideation, executable workflows, scholarly writing and communication, automatic peer review, and end-to-end research. We compare task construction, evidence sources, evaluators, and scoring procedures to explain the capabilities assessed by different designs. Our synthesis highlights three recurring lessons: output checks, process checks, and human studies provide complementary information; evaluator calibration is specific to the property being assessed; and resource budgets and attempt selection are integral to interpreting performance comparisons. We identify diagnostic evaluation designs and documented gaps in supporting evidence, and translate these comparisons into reporting and audit recommendations for specific evaluation settings. The survey helps readers navigate existing evaluations, select appropriate benchmarks, and design subsequent studies.

Generative models can now synthesize highly realistic images, simultaneously increasing the risks of misinformation and visual forgery. Therefore, detecting AI-generated images becomes more essential, and a reliable detector must generalize to unseen generators and stay robust to unseen perturbations in the wild. Existing detectors are typically trained on either independently collected real and generated images or aligned real-generated pairs designed to mitigate content bias. Building on aligned pairs, recent methods form a mixed view by replacing some patches of the real image with their generated counterparts. However, we find that self-attention lets real and generated patches interact, so the feature of each patch no longer reflects its own source alone. This contextual shift makes a per-patch source label an imprecise target. To this end, we propose Relative Patch Response Learning (PRL). Instead of labeling each patch, PRL compares the same patch across two mixed views of an aligned pair and learns from its patch response, the change of its score between the views. (i) To give precise supervision under the contextual shift, a relative response objective measures the responses of source-changed patches against those of source-unchanged patches, which respond to the shift alone. (ii) To provide a reliable reference for the shift, a reference coherence objective keeps each group of source-unchanged patches moving as a whole. (iii) Since the two views contain different amounts of generated content, an area ranking objective asks the view with the larger generated area to have a higher mean patch score. Extensive experiments demonstrate the superior performance of PRL, which surpasses the best prior methods by 4.3% and 5.9% in average balanced accuracy across eight standard and three in-the-wild benchmarks, respectively.

Commercial on-device network operating systems (NOSes) run the complex control planes of production routers and switches, where configuration update tasks are executed by multiple loosely coupled components through asynchronous message passing. Such executions are prone to race conditions: the same ordered input commands may produce different outcomes when messages are delivered in different orders. The race conditions are difficult to expose because they often manifest only as subtle, delayed malfunctions. Existing static analysis techniques lack scalability and precision for large-scale industrial NOSes, while dynamic ones incur substantial system-execution cost when attempting to cover the space of asynchronous message interleavings. To address the challenges above, we present NosRacer, a dynamic analysis framework for race condition detection in industrial-grade on-device NOSes. NosRacer uses a two-phase design. The concentration phase reduces analysis scope by leveraging the locality of configuration update tasks. Race condition detection is then limited to a small subset of involved components and crucial state variables, thereby reducing the detection cost. The perturbation phase proactively perturbs message asynchrony to increase the likelihood of executions with race-condition-exposing message interleavings. It keeps perturbation practical by grouping compatible perturbations for parallel execution and adaptively strengthening perturbations. We implement NosRacer in a commercial, actively developed NOS. NosRacer is integrated into the existing testing factory and used to detect race conditions in 5 key control-plane components. NosRacer achieves 66% precision and detects 21 race conditions confirmed by developers as severe bugs, while keeping the detection overhead below 10%.

Secure deployment of FPGA bitstreams in heterogeneous multi-vendor cloud infrastructures requires scalable authorization, authenticated device access, and bitstream protection throughout the deployment lifecycle. Existing approaches typically address these functions through separate mechanisms, increasing coordination and key management requirements. This paper presents SAEID, a secure FPGA deployment framework that integrates aggregate authorization, certificate-free identity-based device authentication, and identity bound bitstream verification within a pairing based cryptographic framework, with symmetric cryptography used for session binding and AES 256 GCM based bitstream protection. Building on AgEID, an aggregate encryption scheme that enables individual decryption for authorized FPGA devices, SAEID extends the underlying framework to heterogeneous multi vendor deployments by supporting multiple FPGA vendors and IP providers, capability-aware authorization, and dynamic device membership. SAEID provides protection against unauthorized access to future deployments after device revocation and to prior deployments by newly enrolled devices, while retaining constant-size aggregate ciphertexts within each vendor domain and individual decryption. Experimental results demonstrate the practical performance of SAEID, with identity-based device authentication completing in approximately 4.35 ms and device membership updates requiring 4.95 to 5.09 micro seconds for device addition. The aggregate-encryption component exhibits scalable behavior with increasing device-set size. The complete SAEID software decryption path was also validated on a physical ZC702 Cortex A9 platform, requiring 191.126 ms. These results demonstrate scalable aggregate authorization with bounded authentication and dynamic-membership overhead for secure multi-vendor FPGA deployment.

Selection is a task that chooses one element from a finite public candidate range to maximize a data-dependent score. In differential privacy (DP), the exponential mechanism (EM) samples a candidate at a temperature calibrated to the global sensitivity. In this paper, we study when dataset-dependent sensitivity can safely replace global sensitivity in private selection. We propose three valid approaches. First, a private, high-probability upper bound on local sensitivity yields approximate DP, and the method extends to finite higher-order sensitivity hierarchies. Second, our Propose-Test-Release (PTR) variant privately searches a finite public grid for a temperature scale rather than fixing it in advance. Third, smooth sensitivity supports several designs. A candidate-independent smooth geometric construction produces a sensitivity envelope that is admissible under the local dampening framework, which privacy is guaranteed for any admissible envelope. Additionally, a separate logarithmic transformation utilizes smooth sensitivity to produce a smoothed candidate score function with advantages: having controlled global sensitivity, and preserving the maximizers of the original utility score, i.e., candidates maximizing the utility. Both of the designs yield range-independent pure DP. Besides that, we also give two approximate DP private selectors using smooth sensitivity: a direct EM with smooth sensitivity calibrated to the candidate range and privacy parameters that matches a theoretical lower bound up to some constant factor, and one using a privatized smooth upper scale by analyzing the logarithmic transform of the smoothness. For every proposed mechanism, we derive a high-probability regret bound under its stated conditions.

We consider the problem of learning structured linear dynamical systems over convex sets $\mathcal{K}$, where only a small subset of the observations are available at each time point. An estimator which minimizes a bias-corrected, potentially non-convex objective function is proposed. Non-asymptotic bounds are obtained for the statistical error, which depend on the local complexity of $\mathcal{K}$, the trajectory length $T$, and the sub-sampling probability $p$. Convergence of the projected gradient descent algorithm is also established. The general theory is applied to settings where (i) $\mathcal{K}$ is a subspace, (ii) $\mathcal{K}$ is the set of bi-isotonic matrices, and (iii) $\mathcal{K}$ is the set of matrices whose rows are formed by sampling Lipschitz functions. We show meaningful recovery of the transition matrix is possible for values of $T$ much smaller than what is required in the unconstrained case, and for $p = o(1)$.

In deep learning, approximation theory motivates increasing representation size. We ask whether this benefit extends to dynamics learning through autoregressive prediction. We analyze the learned time evolution through the eigenstructure of Koopman operators, using relative residuals to detect spurious eigenpairs arising even as one-step error falls. For bounded Koopman operators, we show that minimal residuals over learned dictionary spaces converge pointwise to their full-space counterparts as these spaces approximate the observable space in $L^2$. Our hypothesis is that Koopman spectral reliability helps explain how consistently rollout error decreases with increasing dimension. We compare two models of a shared Koopman autoencoder trained alternately for reconstruction and latent evolution, using latent-prediction loss (one-step prediction errors in latent coordinates) or spectral-residual loss (relative residuals of candidate eigenpairs). Across six chaotic systems, both models reduced median windowed rollout error from smallest to largest dimension. The spectral-residual model achieved lower medians than the latent-prediction model for all systems and dimensions, and its median fell by a larger factor in every system. Its median decreased monotonically with dimension in four systems, against one for latent prediction. Against four baseline families, its mean valid prediction times were nearly always longer. At the largest dimension under two-stage training, we compared eigenvalue positions with each learned dictionary's residual contours. Spectral-residual eigenvalues concentrated in low-residual regions, whereas latent-prediction eigenvalues also appeared in high-residual regions, consistent with the hypothesis.

Kernel Stein discrepancies (KSDs) provide a versatile tool for comparing distributions. One of their main applications is in quantifying the goodness-of-fit (GoF) between a data-generating distribution and a prescribed target distribution. In this work, we study the related problem of conditional GoF quantification: given only a (possibly non-normalized) conditional target model, without information on the distribution of its covariates, and samples from a joint distribution, the goal is to assess how well the conditional distribution of the samples matches the target. To tackle this setting, we present a framework that allows lifting unconditional KSDs to the conditional setting through an operator-valued kernel on the covariate space, going beyond the known Euclidean case. We establish that our suggested statistic vanishes if and only if the conditional model and the true conditional distribution agree for almost all covariates and deploy it to test conditional GoF on smooth manifolds and on discrete spaces. Our experiments on level, power, and runtime demonstrate the viability of testing on these domains using the proposed statistic.

Mathematicians value proofs for more than correctness: among correct proofs, simplicity, purity and the computational cost of finding them vary widely. Yet LLM-powered theorem provers largely search for any correct proof, and improve its quality only after it is found. We propose LEVER, a proof search algorithm that makes the objective over correct proofs programmable and optimizes it during search. LEVER scores partial proofs over an AND/OR proof graph, combining realized objective values with predictions for open subgoals, so the objective guides search before a proof is complete. The same mechanism optimizes computational cost, proof length, topical impurity, and even their weighted combinations, while the Lean kernel enforces correctness. On PutnamBench in Lean 4, under matched budgets, LEVER costs 34% less than a strong single-conversation agent while raising the solve rate from 80% to 96%. On reducing topical impurity, i.e., how far a proof strays from its theorem's subject, it improves over post-hoc refactoring (42% reduction against 33%) at two-thirds of the cost and more reliably; on proof length, the metric refactoring is built for, it approaches refactoring. Varying the objective's weights traces a quality-cost trade-off curve, so the user can choose how much a better proof is worth. Overall, LEVER is a performant, cost-efficient and tunable proof search algorithm for navigating the space of correct proofs.

Agent skills provide reusable guidance for code agents, but incomplete or unsuitable guidance can impair task execution. To reduce the manual effort of skill refinement, recent approaches use LLMs to generate revisions from execution feedback. However, grounding these revisions in explicit behavioral evidence remains challenging. To address this gap, we propose SkillMorph, a skill-evolution approach based on trajectory-guided fault localization in agent skills. Its core idea is to link execution evidence to specific skill contents before generating revisions. Specifically, SkillMorph compares failure and success evidence in abstracted trajectories across repeated runs and tasks, incorporating changes between evolution loops to identify suspicious actions. It then uses these suspicious actions to localize edit sites in the skills and generate corresponding revisions. Experiments on SWE-Skills-Bench and CannBot show that the skills evolved by SkillMorph consistently achieve higher trial-level accuracy and execution consistency than the original skills and those from four existing skill-evolution methods. We have also applied SkillMorph to automated kernel generation with an AI operator-development team, which has accepted 6 skill-revision pull requests.

3D scene inpainting aims to recover missing or occluded regions in edited 3D scenes, while ensuring geometric and textural consistency. Existing approaches, however, typically require accurately calibrated camera poses, which restricts their applicability in casual, in-the-wild scenarios and introduces additional preprocessing overhead. To overcome this limitation, we present FreeInpaint, a novel feed-forward framework that generates complete and 3D-consistent scenes directly from unposed multi-view images with masked regions. At its core, FreeInpaint extends a 3D foundation model to propagate masked regions from a reference view to other unposed views, bridging 3D reconstruction and scene inpainting while preserving the model's native ability to recover camera poses and scene geometry. Our method addresses two key challenges in adapting feed-forward 3D foundation models to masked inputs. First, masked regions can corrupt cross-view correspondence reasoning, degrading pose estimation and geometry recovery. To address this, we introduce a Learnable Mask Attention mechanism that preserves the spatial anchoring of reliable observations while allowing masked regions to progressively absorb useful context in deeper layers. Second, under severe occlusions, a single forward pass often lacks sufficient appearance evidence for high-fidelity completion. Therefore, we propose a Support Token Refinement strategy, which injects diffusion-generated support evidence as confidence-weighted auxiliary tokens to refine under-observed regions while preserving the original spatial anchor. Extensive experiments across diverse datasets demonstrate that FreeInpaint achieves superior inpainting quality, eliminating the reliance on pre-computed camera poses while keeping a fast inference speed. The project page is https://rorisis.github.io/FreeInpaint/.

Reinforcement learning (RL) methods such as GRPO substantially improve large language model reasoning but often suffer from policy entropy collapse: the loss of sampling diversity weakens exploration and limits further improvement. Existing methods address this issue either through algorithm-level interventions, such as reward modification and entropy/KL regularization, or through token-level reweighting. We investigate a complementary perspective: entropy collapse can also be mitigated by changing which generated rollouts contribute to policy updates. Under the same sampling budget, not all rollouts contribute positively to an update, and selectively excluding some can improve learning. To address this, we propose GRPODropout: before the standard update, we use a simple strategy that selectively removes a small number of high-probability positive-advantage rollouts and recenters the retained advantages. To motivate this design, we develop a rollout-level theoretical analysis that guides method design and threshold selection. The method changes only rollout usage, and adds negligible computational overhead. Experiments show higher accuracy than original GRPO and higher actor entropy while using fewer rollout samples for updates, illustrating "less is more." This work provides insight into RL rollout usage: removing some rollouts can improve performance. Code is available at https://github.com/hexuandeng/GRPODropout/.

Modern neural networks are heavily over-parameterized. This redundancy incurs substantial compute and memory overhead during training and inference. Existing pruning methods rely on post-hoc magnitude thresholds or static initialization heuristics. Consequently, they often require manual per-layer sparsity targets or expensive retraining cycles. We propose Dynamic Activity-Dependent Pruning (DADP), a biologically inspired structural plasticity mechanism. During training, DADP measures connection importance via the accumulated product of pre-synaptic activations and post-synaptic error gradients. Using a single global threshold instead of fixed layer budgets, DADP dynamically allocates sparsity across network depth while naturally inducing neuron- and channel-level pruning. Across MLP, VGG-16, ResNet-18, BiLSTM-CRF, and MiniBERT architectures, DADP matches or outperforms Magnitude, SNIP and RigL, retaining 73.67% accuracy (dense baseline: 76.06%) at 99% sparsity on ResNet-18. Finally, matrix-based Shannon entropy and effective rank measurements confirm that DADP preserves latent feature diversity at extreme sparsities without representation collapse.

The Hydrogen Plasma Smelting Reduction (HPSR) is a promising technology for low-CO$_2$ steel production. However, unstable arc conditions limit energy efficiency, hydrogen utilization, and electrode wear. In this paper, we discuss the influence of the electrical supply characteristics, which can be a source of additional instability. For active stabilization, a fast current controller using a buck converter with an adaptive state controller is used. For analyzing arc stability, a stereoscopic measurement system for reconstructing the arc is presented. The results show a clear correlation between the arc lengths and the measured arc voltage at constant current. Therefore, arc stability can be determined by the measured voltage. These findings build the basis for dynamic arc control by fast electrical interventions.

Finding the right video often requires distinguishing similar scenes in which different events occur. Joint matching improves retrieval, but processing rich video representations for each query is costly. VEDJE compresses features within sampled frames while keeping their representations separate in a reusable cache. Feature-change prediction supplies an auxiliary training signal that improves retrieval from the compressed cache without adding work at query time. On MSR-VTT, MSVD, DiDeMo, and ActivityNet, VEDJE improves R@1 over matched first-stage retrievers in both retrieval directions. On MSR-VTT, it reaches 59.8 text-to-video R@1 with a fine-tuned VideoCLIP-XL first stage. In the VideoPrism configuration, shrinking the per-video cache fourfold to 12 KiB preserves text-to-video recall within 0.2 points. These results show that accurate video search can operate on compact evidence, encoded once and reused as new queries arrive.

The rapid proliferation of Large Language Models (LLMs) necessitates reliable watermarking techniques to identify AI-generated text and ensure appropriate attribution. While token-based watermarks are vulnerable to paraphrasing, a central challenge for semantic watermarking is to turn sentence meanings into a stable, well-calibrated document-level signal. To this end, we introduce SemField, a simple, training-free semantic watermark that embeds a continuous, linear signal directly into the sentence embedding space. Using a shared secret key to define a specific Gaussian direction, SemField iteratively evaluates candidate sentences and selects those that maximize the alignment of the cumulative document aggregate with this targeted direction. We also propose SemField-PL, a polarity-locked variant that first determines the optimal orientation from the initial sentence and continuously reinforces it throughout the generation process. The document-level aggregated formulation guarantees exact invariance to sentence reordering and provides theoretical bounds against structural tampering, such as sentence insertion and deletion. Lastly, with extensive evaluations across three models, we demonstrate that both variants outperform 12 recent baselines. Across clean detection and four paraphrasing attacks, both variants achieve a mean True Positive Rate (TPR) of 88.2% to 90.4% at a 1% False Positive Rate (FPR), all while maintaining comparable perplexity and naturalness as human-generated content. We open-source our implementation at https://github.com/declare-lab/SemField