arXiv subjects

Lei Yang

Publications and source records attributed to Lei Yang.

At least 55 records · Page 3Linked to original sources

arXiv2026

AdaptiveEmbed: Sample-Adaptive Multi-Vector Representation for Multimodal Retrieval

Multi-vector representations have emerged as an effective paradigm for multimodal retrieval, representing each sample with multiple complementary embeddings to capture fine-grained cross-modal information. However, existing approaches typically employ a fixed representation capacity, assigning the same number of vectors to all samples regardless of their individual retrieval demands. Such a fixed-capacity formulation overlooks the fact that different samples may require different amounts of representation capacity for effective retrieval. In this work, we introduce \emph{Sample-Adaptive Multi-Vector Representation} (SAMVR), a new problem setting for multimodal retrieval that studies how multi-vector representation capacity can be allocated at the sample level. Under SAMVR, each sample is represented by a \emph{content-adaptive embedding set} (CAES), whose capacity is determined according to the sample-specific retrieval utility of additional representation vectors. To instantiate SAMVR, we propose \emph{AdaptiveEmbed}, a unified framework for learning sample-adaptive multi-vector representations. AdaptiveEmbed learns structured multi-vector representations through \emph{Multi-Group Contrastive Learning} (MGCL) with the symmetric \emph{set-to-set similarity} (SetSim), and further employs \emph{Utility Policy Optimization} (UPO) to determine sample-specific representation capacity via \emph{Marginal Utility Allocation} (MUA). Experiments across multimodal retrieval benchmarks involving image, text, video, and audio show that sample-adaptive capacity allocation achieves overall better retrieval performance than fixed-capacity multi-vector representations, validating the effectiveness of SAMVR for multimodal retrieval. These results establish SAMVR as a viable formulation for adaptive capacity allocation in multi-vector multimodal retrieval.

cs.CV↗

arXiv2026

GaussianDream++: Efficient 3D Gaussian World Modeling for Robotic Manipulation

Vision-Language-Action (VLA) policies have advanced language-conditioned robotic manipulation, yet action-imitation objectives provide only weak supervision for metric 3D structure and short-horizon physical evolution. Geometry-enhanced policies mainly improve current-scene grounding, whereas predictive policies often model future dynamics in RGB or latent spaces and may incur substantial deployment cost. GaussianDream demonstrates that training-time current Gaussian reconstruction and future Gaussian prediction provide effective 3D supervision, but its dense VGGT/TGE-based prefix jointly carries state, dynamics, and action-conditioning information. We present \textbf{\methodname}, a compact, policy-native extension that inserts \textbf{World State Tokens} and \textbf{World Prediction Tokens} directly into the VLA backbone. A training-only \textbf{World Representation Head} decodes these tokens into a Current World and coupled Future Prediction over shared Gaussian primitives, while static--dynamic factorization preserves persistent structure and focuses residual motion on interaction-relevant regions. At inference, the head, renderer, auxiliary objectives, and VGGT/TGE pathway are removed, leaving only 20 world tokens without online Gaussian decoding or rollout. \method achieves \textbf{98.6\%} on LIBERO and \textbf{87.8\%} on LIBERO-Plus, with clear gains under Camera and Layout shifts. Real-robot experiments further improve average success from 29.2\% to 52.5\% over reproduced $π_{0.5}$ while maintaining efficient closed-loop control.

cs.RO↗

arXiv2026

TransPhy: Visual In-Context Learning for Physically Grounded Image Editing

Visual demonstrations provide a natural interface for specifying image transformations that are difficult to describe exhaustively with text. However, existing visual in-context learning (VICL) methods primarily focus on appearance-level relation transfer and provide limited support for physically grounded transformations, whose outcomes depend on material properties, geometry, object interactions, and environmental conditions. Given a source--target exemplar pair and a query image, physically grounded VICL requires a model to infer the demonstrated transformation, adapt its effects to the query-specific scene context, and preserve rule-irrelevant content. We introduce PhysVICL-74, comprising 74 physically grounded transformation rules and 5,240 source--target image pairs that form nearly 75K training and evaluation contexts. Its benchmark split separately evaluates novel-instance transfer and unseen-rule generalization. We further propose TransPhy, a framework that decomposes physically grounded VICL into physical-rule induction and transition-aligned rendering. TransPhy first predicts the demonstrated rule and an explicit query-specific target-state description, and then synthesizes the target image through token-wise mixture-of-experts adaptation, with expert routing guided by localized transition cues. Experiments show that TransPhy improves physical-rule adherence, query consistency, and unseen-rule generalization over existing visual in-context editing methods.

cs.CV↗

arXiv2026

Not All History Helps: Velocity-Aware Selective Memory for Long-Horizon End-to-End Autonomous Driving

Reliable long-horizon planning remains a key challenge in end-to-end autonomous driving. By accounting for future motion evolution and potential consequences, it provides forward-looking guidance for safe and consistent driving in evolving traffic environments. Existing methods use historical planning states as temporal context. Self-generated history may become stale or conflict with the current motion stage, introducing unreliable priors. We propose StableDrive to address cross-cycle historical reliability and within-horizon motion-stage evolution. Selective Momentum Memory (SMM), implemented with a Mamba selective state-space operator, controls the influence of the preceding self-predicted planning state on the current cycle. Motion-Stage Training Scaffold (MSTS) uses motion-stage, long-horizon trajectory, and longitudinal-motion supervision to guide stage-aware future motion learning and is removed before inference. A fixed parameter midpoint between two architecture-aligned endpoints yields a single deployable SMM planner without model ensembling or extra inference-time computation. On nuScenes under the MomAD evaluation protocol, StableDrive achieves SOTA performance across all reported planning metrics from 1 to 6 s, reducing average collision rate by 23.3%, TPC by 30.9%, and L2 by 11.8% over the best previously reported value for each metric. On the curated Longitudinal-Transition nuScenes (LT-nuScenes), StableDrive reduces 6-s collision rate by 23.81%, TPC by 10.90%, and L2 by 6.37%. On NAVSIM v1 and v2, StableDrive achieves the highest PDMS/EPDMS in all three reported settings, including a 5.7-point EPDMS gain on v2 navhard over the previous best.

cs.RO↗

arXiv2026

SPARGen: Unifying Spatial Perception and Reasoning through Native Multimodal Generation

Spatial perception and reasoning from visual observations require recovering geometric structure, establishing correspondences, and understanding spatial relations. Existing approaches typically address these capabilities separately using task-specific architectures or external geometric modules, limiting knowledge transfer among complementary representations of the same physical scene. We introduce SPARGen, a unified multimodal framework that casts 3D reconstruction, dense correspondence, and spatial reasoning as instruction-conditioned generation tasks. SPARGen serializes compact structured and linguistic outputs as token sequences while generating dense geometric fields in image-aligned forms, enabling spatial supervision to jointly shape shared representations within a native multimodal generative model. Experiments across benchmarks for 3D reconstruction, correspondence, and spatial reasoning show that SPARGen achieves competitive performance across heterogeneous spatial tasks within a single native multimodal generative framework.

cs.CV↗

arXiv2026

A Hybrid Statistical Learning Framework for Capability-Based Decision Support in Manufacturing

Process capability indices are widely used in manufacturing quality control, but capability approval is often implemented by directly thresholding finite-sample estimates, which can produce unstable and poorly calibrated decisions near the approval boundary. This paper develops a hybrid statistical--learning framework for capability-based decision support in manufacturing. The proposed UC-Cap approach combines a statistically grounded capability baseline with a residual learning component that uses process, distributional, specification-related, and measurement-related features to refine capability-decision risk estimates under non-ideal manufacturing conditions. The statistical baseline preserves the interpretability of classical capability analysis, while the learning component provides data-driven correction for systematic deviations arising from non-normality, measurement effects, and finite-sample variability. A nested Monte Carlo evaluation is introduced to assess probabilistic calibration under controlled synthetic settings, and an empirical manufacturing study is used to examine decision behavior under realistic capability data. Results show that deterministic thresholding can lead to substantial instability and miscalibration in near-boundary regimes, whereas the proposed framework provides calibrated risk estimates, interpretable decision outputs, and improved support for capability approval decisions. The framework is compatible with existing capability-analysis workflows and can be integrated into manufacturing quality decision-support systems.

stat.AP↗

arXiv2026

Demystifying Video Reasoning

Recent advances in video generation have revealed an unexpected phenomenon: diffusion-based video models exhibit non-trivial reasoning capabilities. Prior work attributes this to a Chain-of-Frames (CoF) mechanism, where reasoning is assumed to unfold sequentially across video frames. In this work, we challenge this assumption and uncover a fundamentally different mechanism. We show that reasoning in video models instead primarily emerges along the diffusion denoising steps. Through qualitative analysis and targeted probing experiments, we find that models explore multiple candidate solutions in early denoising steps and progressively converge to a final answer, a process we term Chain-of-Steps (CoS). Beyond this core mechanism, we identify several emergent reasoning behaviors critical to model performance: (1) working memory that supports tasks requiring consistent reference, such as object permanence; (2) self-correction and enhancement, allowing recovery from incorrect intermediate solutions; and (3) perception before action, where early steps establish semantic grounding and later steps perform structured manipulation. Moreover, analysis of Diffusion Transformer layers shows that middle layers conduct key reasoning procedures. Motivated by these insights, we present a simple Training-Free Ensemble (TFE) as a proof-of-concept, demonstrating how reasoning can be improved by ensembling latent trajectories from identical models with different random seeds. Overall, our work provides the first systematic dissection of the mechanisms underlying video reasoning, offering a foundation to guide future research in better exploiting the inherent reasoning dynamics of video models as a new substrate for intelligence.

cs.CV↗

arXiv2026

Newton Method for Fixed-Support Doubly Entropic Wasserstein Barycenter

We study the fixed-support doubly regularized Wasserstein barycenter problem. Using the semi-dual formulation of entropic optimal transport, we reformulate the problem as a smooth, unconstrained, convex optimization problem in the dual variables. We then derive explicit expressions for the gradient and Hessian and develop an exact Newton method for high-accuracy barycenter computation. To improve scalability, we propose a sparse Newton variant that sparsifies the transport probability matrices, thereby reducing the cost of Hessian-vector products. We establish theoretical results for the proposed methods, including Hessian approximation bounds and convergence results. Experiments on synthetic and real datasets show that the sparse Newton method converges faster than

math.OC↗

arXiv2026

Hallucinations Leave a Grounding Signature:Verifier-Guided Decoding for Selective Object Correction

Large vision-language models (LVLMs) often hallucinate objects that are absent from an image. Despite recent progress, existing mitigation methods still lack reliable object-level grounding diagnostics and therefore tend to apply coarse-grained interventions, which can impair visual understanding, shorten responses, and reduce coverage of genuinely grounded objects. The key challenge is thus to detect, during generation, whether each emerging object mention is supported by reliable visual evidence, so that hallucination can be mitigated selectively. Yet output confidence reflects next-token plausibility rather than visual support, allowing language priors to make absent objects appear certain. We show that the missing diagnostic evidence is encoded in an Intrinsic Grounding Signature (IGS), a distributed signed attention pattern that remains informative for such confident hallucinations. Based on IGS, we propose Verifier-Guided Decoding (VGD), a decoding framework in which a lightweight verifier examines each emerging object mention, rolls back the KV cache when the mention is identified as high risk, suppresses the object and its synonyms, and regenerates the affected continuation. Because VGD intervenes only on object mentions identified as high risk, it reduces object hallucination while preserving the model's original visual understanding and grounded object coverage. Experiments on CHAIR and AMBER-G show that VGD achieves state-of-the-art object hallucination reduction: at @rec90, it cuts AMBER-G CHAIR by 43.6\% while retaining 99.6\% of grounded-object coverage, and reduces CHAIR-MSCOCO CHAIR$_i$/CHAIR$_s$ by 37.0\%/30.4\% without shortening captions.

cs.CV↗

arXiv2026

Deblur-Avatar: Animatable Avatars from Motion-Blurred Monocular Videos

We introduce a novel framework for modeling high-fidelity, animatable 3D human avatars from motion-blurred monocular video inputs. Motion blur is prevalent in real-world dynamic video capture, especially due to human movements in 3D human avatar modeling. Existing methods either (1) assume sharp image inputs, failing to address the detail loss introduced by motion blur, or (2) mainly consider blur by camera movements, neglecting the human motion blur which is more common in animatable avatars. Our proposed approach integrates a human movement-based motion blur model into 3D Gaussian Splatting (3DGS). By explicitly modeling human motion trajectories during exposure time, we jointly optimize the trajectories and 3D Gaussians to reconstruct sharp, high-quality human avatars. We employ a pose-dependent fusion mechanism to distinguish moving body regions, optimizing both blurred and sharp areas effectively. Extensive experiments on synthetic and real-world datasets demonstrate that our method significantly outperforms existing methods in rendering quality and quantitative metrics, producing sharp avatar reconstructions and enabling real-time rendering under challenging motion blur conditions. Code and models are available at https://github.com/xianrui-luo/deblur_avatar.

cs.CV↗

arXiv2026

SGMD: Score Gradient Matching Distillation for Few-Step Video Diffusion Distillation

Distribution Matching Distillation (DMD) is a widely used paradigm for accelerating inference in few-step video diffusion models. However, DMD-style video distillation faces two coupled challenges: the fake score must track a continuously evolving generator, making training costly when frequent updates are required, while reverse-KL-style matching can be mode-seeking and conservative for preserving strong motion dynamics. To address these issues, we propose \textbf{Score Gradient Matching Distillation (SGMD)}. SGMD adopts a fake-score perspective by directly optimizing the fake score toward the teacher, while using teacher stop-gradient Fisher as a stable distribution-matching objective. We provide a gradient analysis that motivates this objective choice under ideal tracking. Building on this, SGMD introduces a pair of dual potentials: negative-residual (NR) for outer-loop correction and residual-contraction (RC) for inner-loop tracking. Empirically, compared to DMD2, SGMD achieves an approximately $\sim 3\times$ training speedup and substantially improves motion dynamics for 4-step distilled models while preserving temporal consistency. A human study confirms that SGMD is preferred in motion quality and overall preference, while visual quality and text alignment remain comparable. Code is available at https://github.com/ModelTC/LightX2V.

cs.CV↗

arXiv2026

An Inexact Variable Metric Proximal Gradient-subgradient Algorithm for a Class of Fractional Optimization Problems

In this paper, we study a class of fractional optimization problems, in which the numerator of the objective is the sum of a convex function and a differentiable function with a Lipschitz continuous gradient, while the denominator is a nonsmooth convex function. This model captures ratio-type formulations arising in scale-invariant sparse learning and related applications. To address this class of problems, we propose an inexact variable metric proximal gradient-subgradient algorithm (iVPGSA), which, to the best of our knowledge, is the first inexact proximal algorithm specifically designed for such type of fractional problems. By incorporating a variable metric proximal term and allowing for approximate subproblem solutions under a flexible error criterion, the proposed algorithm is highly adaptable to a broader range of problems while achieving favorable computational efficiency. Under suitable assumptions, we establish that any accumulation point of the generated sequence is a critical point of the target problem. Moreover, we develop a new Kurdyka-Łojasiewicz (KL)-based analysis framework, relying only on the classical KL property and its associated exponent, to prove the global convergence of the entire sequence and characterize its convergence rate, \textit{without} requiring a strict sufficient descent property. Our results clarify how the classical KL exponent and inexactness jointly influence the convergence rate. Finally, numerical experiments on the $\ell_1/\ell_2$ Lasso problem and the constrained $\ell_1/\ell_2$ sparse optimization problem demonstrate the computational advantages of the iVPGSA over existing representative algorithms.

math.OC↗

arXiv2026

Apple-$π$: Benchmarking Thinking with Video Towards Law-Grounded Physical Intelligence

Modern video generation models are increasingly hailed as emerging world models with an internalized grasp of physical law. Yet existing benchmarks largely evaluate physical plausibility only at the output level, without verifying whether the model arrives there through a faithful, law-grounded reasoning process. We introduce Apple-PI, the first benchmark that anchors video-model evaluation explicitly in physical laws. Apple-PI comprises three components. 1) Orchard: a dataset of 400 videos covering ten canonical tasks in classical mechanics. It separates single-law tasks for confounder-free diagnosis from multi-law tasks for probing generalization. 2) Benchmark Protocol: a three-stage protocol based on scientific reasoning, including Perception, Formulation, and Deduction. It uses chain-of-frames prompting on infographic-annotated first frames, treating the generated video as the model's visible reasoning trace. 3) Evaluation Suite: a hybrid evaluation suite that combines MLLM-based subjective scoring with physics-law-grounded objective measures. This enables stage-resolved diagnosis of not only whether a model fails, but where it fails. Benchmarking 11 models shows that current video models remain far from reliable law-grounded world simulators, with the best video model scoring only 0.473. Our stage-, pillar-, and source-resolved analyses further expose a Perception-to-Formulation-to-Deduction bottleneck, weak multi-law state transfer, and a persistent Sim-to-Real gap. These findings position Apple-PI as a diagnostic foundation for guiding future video models toward world models with law-grounded physical intelligence.

cs.CV↗

arXiv2026

Moment-Resolved Readout and Reservoir Diversity in Nonequilibrium Langevin Computing

Nonlinear thermodynamic computers based on Langevin dynamics exploit thermal fluctuations as a physical substrate for computation. Recent work has shown that quartic-confined fluctuating degrees of freedom can act as thermodynamic neurons capable of nonlinear function approximation at finite observation times. Here we extend this paradigm from mean-only readout to moment-resolved readout. Instead of representing each driven reservoir solely by its first moment, we construct a response vector from the elementwise raw polynomial moments $\mathbb{E}[\bm{x}]$, $\mathbb{E}[\bm{x}^{\odot 2}]$, and $\mathbb{E}[\bm{x}^{\odot 4}]$. These observables combine displacement and central-shape contributions and are naturally aligned with the linear, quadratic, and quartic terms of the local driven dynamics. We further introduce a heterogeneous multi-reservoir architecture in which three reservoirs with distinct initialization and training histories form a joint $2304$-dimensional response representation. Under the fixed MNIST $60000/10000$ reproduction protocol, feature-level fusion achieves the best observed accuracy of $9695/10000=96.95\%$, compared with $9682/10000=96.82\%$ for the strongest single-reservoir model and $9684/10000=96.84\%$ for equal-weight logit averaging. An exact paired McNemar test does not establish a statistically significant improvement over the strongest single reservoir, but the ablation and wrong-set overlap results provide suggestive evidence of complementary classification errors. These results motivate higher-order polynomial-moment readout and reservoir heterogeneity as candidate design principles for finite-time Langevin computing.

cond-mat.stat-mech↗

arXiv2026

CASA-SDF: Curriculum-Aware Spatial Adaptation with Curvature-Guided Density for Neural Implicit Surface Reconstruction

Neural implicit representations have emerged as a powerful paradigm for 3D reconstruction. However, high-fidelity indoor surface reconstruction remains a significant challenge, primarily due to the pronounced \emph{geometric heterogeneity} of indoor scenes. Large texture-less planar regions typically require stronger regularization to suppress high-frequency artifacts, while thin structures demand sharper, more adaptive representations to mitigate the spectral bias of multi-layer perceptrons (MLPs) and prevent over-smoothing. Existing approaches often rely on spatially indiscriminate prior supervision and a scene-global SDF-to-density transformation, which constrains their ability to balance planar smoothness and detail preservation. In this paper, we propose CASA-SDF (Curriculum-Aware Spatial Adaptation for SDF), a unified framework that addresses this challenge via complementary adaptations of supervision and representation capacity. Specifically, Hybrid Spatially-Adaptive Uncertainty Annealing (SAUA) fuses semantic and photometric uncertainties to construct a pixel-wise curriculum for monocular prior supervision. This strategy maintains regularization in reliable regions while attenuating unreliable supervision early in training to enable data-driven photometric refinement. Meanwhile, Curvature-Aware Locally Adaptive Density Transformation (CALADT) progressively modulates the sharpness of the SDF-to-density mapping via a curvature proxy to enhance the representation of thin structures. Extensive experiments on benchmark indoor datasets demonstrate that CASA-SDF improves surface completeness and detail recovery on high-frequency structures, without compromising the stability of planar surfaces.

cs.CV↗

arXiv2026

Radio and X-ray flux rebrightening six years after outburst in a partially-obscured extreme changing-look AGN

SDSS J1548+2208 is a unique partially-obscured nuclear transient that exhibits multiwavelength outbursts in mid-infrared, X-ray and radio. We present the results from multiwavelength photometric and spectroscopic follow-up observations with a time span of ~2500 days since its discovery. We find that the mid-infrared and X-ray emission (with a hard X-ray spectrum) are still in a high flux level relative to the pre-flare state, suggesting a sudden increased, and possibly long-sustained accreting activity from central black hole. This is supported by the slowly-evolving high-ionization coronal lines. The mid-infrared color turns blue slowly in the rising phase, which is distinct from stellar tidal disruption events (TDEs). All these properties point to the origin of outbursts from an extreme changing-look AGN and the scenario with a normal TDE seems disfavored. The radio spectral energy distribution (SED) in ~0.65-15 GHz is unusual, displaying a double-peak feature with distinct variability characteristics. In addition, we find evidence for the late-time radio rebrightening more than six years since the initial outburst, as well as a possibly new X-ray flare, though the significance for the latter is not high. The peculiar radio flux and SED evolution could be explained by a nascent outflow expanding into and shocking circumnuclear diffuse medium filled by denser clouds. In this case, SDSS J1548+2208 represents a rare changing-look AGN which can launch radio outflows. Continued multiwavelength observations are required to map the dust and gas distribution on pc-scales, providing new insights into the environmental properties that could regulate AGN changing-look phenomenon.

astro-ph.HE↗

arXiv2026

AT2019ijn: a fast-rising, slow-decaying blue optical transient with exceptionally bright radio emission

We report the discovery of a peculiar optical transient, AT2019ijn, occurred in the nuclear region of a dwarf galaxy at z=0.273. It rises rapidly to peak at a luminosity of Mg=-21.1 in 5 days, followed by a slow decline over more than a month, during which the optical emission has a persistently high blackbody temperature of T_BB~1.5-1.6x10^4 K. The radio emission is exceptional which peaks at 640 days after optical discovery with a high luminosity of 2x10^31 erg/s/Hz. The peak radio luminosity is at least two orders of magnitude brighter than known radio-bright fast blue optical transients and supernova explosions at similar epochs, but comparable to jetted tidal disruption events. The luminous and long-lasting radio emission with a late-time peak can be explained by an off-axis relativistic jet with a viewing angle of ~40 deg. We discuss possible origins for AT2019ijn and favor a jetted tidal disruption event involving an intermediate-mass black hole of ~10^5 Msun, although a jetted magnetar model cannot be fully ruled out. AT2019ijn represents a new class of relativistic optical transients that highlights the importance of radio surveys for discovering off-axis jetted events.

astro-ph.HE↗

arXiv2026

MUX-USCT: A Noise-Robust Neural Network for Ultrasound Computed Tomography

Deep neural networks (DNNs) have shown strong potential for ultrasound computed tomography (USCT) reconstruction in ideal noise-free environments, yet existing DNNs are vulnerable to the noisy conditions in clinical practice, as they equally treat inputs that suffer mild, moderate, or severe noise. More challenging, the distributions of noise shift along with the environment, indicating the less effectiveness of noise-aware training, which injects a specific noise distribution into the training data. We rethink these challenges and observe that the DNN models can become more robust to noise if we know the noise sources and filter them out. This filtering operation is very alike the Multiplexers (or MUX), a fundamental combinational circuit in digital logic design. However, the challenge here is that noise can happen randomly during inference; as a result, the manually predefined MUX cannot work. To address these challenges, we propose MUX-USCT, a novel encoder-decoder DNN architecture that encodes the known acoustic acquisition geometry with an "adaptive MUX" that can automatically identify and filter noise, where the attention mechanism is applied in reconstructing the speed-of-sound map. On the OpenPros benchmark, MUX-USCT reaches 6.88 m/s MAE with 17% fewer parameters than the leading baseline with 7.65 m/s of MAE. Under simulated clinical noise, it remains stable across diverse degradation types that cause geometry-agnostic baselines to fail. Results show that the attention distributions in MUX-USCT provide interpretable indicators of the signal quality between pairs of transducers.

eess.IV↗