arXiv subjects

Wei Gao

Publications and source records attributed to Wei Gao.

At least 109 records · Page 6Linked to original sources

arXiv2025

DriveExplorer: Images-Only Decoupled 4D Reconstruction with Progressive Restoration for Driving View Extrapolation

This paper presents an effective solution for view extrapolation in autonomous driving scenarios. Recent approaches focus on generating shifted novel view images from given viewpoints using diffusion models. However, these methods heavily rely on priors such as LiDAR point clouds, 3D bounding boxes, and lane annotations, which demand expensive sensors or labor-intensive labeling, limiting applicability in real-world deployment. In this work, with only images and optional camera poses, we first estimate a global static point cloud and per-frame dynamic point clouds, fusing them into a unified representation. We then employ a deformable 4D Gaussian framework to reconstruct the scene. The initially trained 4D Gaussian model renders degraded and pseudo-images to train a video diffusion model. Subsequently, progressively shifted Gaussian renderings are iteratively refined by the diffusion model,and the enhanced results are incorporated back as training data for 4DGS. This process continues until extrapolation reaches the target viewpoints. Compared with baselines, our method produces higher-quality images at novel extrapolated viewpoints.

cs.CV↗

arXiv2025

Ultrahigh-Energy Gamma-ray Emission Associated with Black Hole-Jet Systems

Black holes (BH), one of the most intriguing objects in the universe, can manifest themselves through electromagnetic radiation initiated by the accretion flow. Some stellar-mass BHs drive relativistic jets when accreting matter from their companion stars, forming microquasars. Non-thermal emission from the radio to tera-electronvolt (TeV) gamma-ray band has been observed from microquasars, indicating the acceleration of relativistic particles. Here we report detection of four microquasars (SS 433, V4641 Sgr, GRS 1915+105, MAXI J1820+070) of spectrum extending to the ultrahigh-energy (UHE; photon energy $E>100$ TeV) band and one microquasar (Cygnus X-1) of spectrum approaching 100 TeV, using the Large High Altitude Air Shower Observatory (LHAASO). Notably, the total emission associated with SS 433 cannot be interpreted with a single leptonic component. In the UHE band, its emission is in spatial coincidence with a giant atomic cloud, which is consistent with a hadronic origin. An elongated source is discovered from V4641 Sgr with the spectrum continuing up to 800 TeV. The detection of UHE gamma rays demonstrates that accreting BHs and their environments can operate as extremely efficient accelerators of particles out of 1 peta-electronvolt (PeV), suggesting microquasars to be important contributors to Galactic cosmic rays especially around the `knee' region.

astro-ph.HE↗

arXiv2025

Joint Link Adaptation and Device Scheduling Approach for URLLC Industrial IoT Network: A DRL-based Method with Bayesian Optimization

In this article, we consider an industrial internet of things (IIoT) network supporting multi-device dynamic ultra-reliable low-latency communication (URLLC) while the channel state information (CSI) is imperfect. A joint link adaptation (LA) and device scheduling (including the order) design is provided, aiming at maximizing the total transmission rate under strict block error rate (BLER) constraints. In particular, a Bayesian optimization (BO) driven Twin Delayed Deep Deterministic Policy Gradient (TD3) method is proposed, which determines the device served order sequence and the corresponding modulation and coding scheme (MCS) adaptively based on the imperfect CSI. Note that the imperfection of CSI, error sample imbalance in URLLC networks, as well as the parameter sensitivity nature of the TD3 algorithm likely diminish the algorithm's convergence speed and reliability. To address such an issue, we proposed a BO based training mechanism for the convergence speed improvement, which provides a more reliable learning direction and sample selection method to track the imbalance sample problem. Via extensive simulations, we show that the proposed algorithm achieves faster convergence and higher sum-rate performance compared to existing solutions.

cs.LG↗

arXiv2025

Explainable Ethical Assessment on Human Behaviors by Generating Conflicting Social Norms

Human behaviors are often guided or constrained by social norms, which are defined as shared, commonsense rules. For example, underlying an action ``\textit{report a witnessed crime}" are social norms that inform our conduct, such as ``\textit{It is expected to be brave to report crimes}''. Current AI systems that assess valence (i.e., support or oppose) of human actions by leveraging large-scale data training not grounded on explicit norms may be difficult to explain, and thus untrustworthy. Emulating human assessors by considering social norms can help AI models better understand and predict valence. While multiple norms come into play, conflicting norms can create tension and directly influence human behavior. For example, when deciding whether to ``\textit{report a witnessed crime}'', one may balance \textit{bravery} against \textit{self-protection}. In this paper, we introduce \textit{ClarityEthic}, a novel ethical assessment approach, to enhance valence prediction and explanation by generating conflicting social norms behind human actions, which strengthens the moral reasoning capabilities of language models by using a contrastive learning strategy. Extensive experiments demonstrate that our method outperforms strong baseline approaches, and human evaluations confirm that the generated social norms provide plausible explanations for the assessment of human behaviors.

cs.CY↗

arXiv2025

RollMux: Phase-Level Multiplexing for Disaggregated RL Post-Training

Rollout-training disaggregation is emerging as the standard architecture for Reinforcement Learning (RL) post-training, where memory-bound rollout and compute-bound training are physically disaggregated onto purpose-built clusters to maximize hardware efficiency. However, the strict synchronization required by on-policy algorithms introduces severe dependency bubbles, forcing one cluster to idle while the dependent phase is running on the other. We present RollMux, a cluster scheduling framework that reclaims these bubbles through cross-cluster orchestration. RollMux is built on the insight that the structural idleness of one job can be effectively utilized by the active phase of another. To realize this, we introduce the co-execution group abstraction, which partitions the cluster into isolated locality domains. This abstraction enables a two-tier scheduling architecture: an inter-group scheduler that optimizes job placement using conservative stochastic planning, and an intra-group scheduler that orchestrates a provably optimal round-robin schedule. The group abstraction also imposes a residency constraint, ensuring that massive model states remain cached in host memory to enable "warm-star" context switching. We evaluate RollMux on a production-scale testbed with 328 H20 and 328 H800 GPUs. RollMux improves cost efficiency by 1.84x over standard disaggregation and 1.38x over state-of-the-art co-located baselines, all while achieving 100% SLO attainment.

cs.DC↗

arXiv2025

Estimation of Treatment Effects based on Kernel Matching

Kernel matching is a widely used technique for estimating treatment effects, particularly valuable in observational studies where randomized controlled trials are not feasible. While kernel-matching approaches have demonstrated practical advantages in exploiting similarities between treated and control units, their theoretical properties have remained only partially explored. In this paper, we make a key contribution by establishing the asymptotic normality and consistency of kernel-matching estimators for both the average treatment effect (ATE) and the average treatment effect on the treated (ATT) through influence function techniques, thereby providing a rigorous theoretical foundation for their use in causal inference. Furthermore, we derive the asymptotic distributions of the ATE and ATT estimators when the propensity score is estimated rather than known, extending the theoretical guarantees to the practically relevant cases. Through extensive Monte Carlo simulations, the estimators exhibit consistently improved performance over standard treatment-effect estimators. We further illustrate the method by analyzing the National Supported Work Demonstration job-training data with the kernel-matching estimator.

stat.ME↗

arXiv2025

A Large-Scale Multimodal Dataset and Benchmarks for Human Activity Scene Understanding and Reasoning

Multimodal human action recognition (HAR) leverages complementary sensors for activity classification. Beyond recognition, recent advances in large language models (LLMs) enable detailed descriptions and causal reasoning, motivating new tasks: human action understanding (HAU) and human action reasoning (HARn). However, most LLMs, especially large vision language models (LVLMs), struggle with non-RGB modalities such as depth, IMU, and mmWave due to the lack of large-scale data-caption resources. Existing HAR datasets mainly provide coarse data-label annotations, which are insufficient to capture fine-grained action dynamics needed for HAU and HARn. We consider two ground-truth pair types: (1) data label (discrete category) and (2) data caption (textual description). Naively generating captions from labels often lacks logical and spatiotemporal consistency. We introduce CUHK-X, a large-scale multimodal dataset and benchmark suite for HAR, HAU, and HARn. CUHK-X contains 58,445 samples covering 40 actions performed by 30 participants across two indoor environments. To improve caption consistency, we propose a prompt-based scene creation method that leverages LLMs to generate logically connected activity sequences, followed by human validation. CUHK-X includes three benchmarks with six evaluation tasks. Experiments report average accuracies of 76.52% (HAR), 40.76% (HAU), and 70.25% (HARn). CUHK-X aims to enable the community to apply and develop data-intensive learning methods for robust, multimodal human activity analysis. Project page and code: https://openaiotlab.github.io/CUHK-X/ and https://github.com/openaiotlab/CUHK-X.

cs.CV↗

arXiv2025

InfiniBench: Infinite Benchmarking for Visual Spatial Reasoning with Customizable Scene Complexity

Modern vision-language models (VLMs) are expected to have abilities of spatial reasoning with diverse scene complexities, but evaluating such abilities is difficult due to the lack of benchmarks that are not only diverse and scalable but also fully customizable. Existing benchmarks offer limited customizability over the scene complexity and are incapable of isolating and analyzing specific VLM failure modes under distinct spatial conditions. To address this gap, instead of individually presenting benchmarks for different scene complexities, in this paper we present InfiniBench, a fully automated, customizable and user-friendly benchmark generator that can synthesize a theoretically infinite variety of 3D scenes with parameterized control on scene complexity. InfiniBench uniquely translates scene descriptions in natural language into photo-realistic videos with complex and physically plausible 3D layouts. This is achieved through three key innovations: 1) a LLM-based agentic framework that iteratively refines procedural scene constraints from scene descriptions; 2) a flexible cluster-based layout optimizer that generates dense and cluttered scenes previously intractable for procedural methods; and 3) a task-aware camera trajectory optimization method that renders scenes into videos with full object coverage as VLM input. Experiments demonstrate that InfiniBench outperforms state-of-the-art procedural and LLM-based 3D generation methods in prompt fidelity and physical plausibility, especially in high-complexity scenarios. We further showcased the usefulness of InfiniBench, by generating benchmarks for representative spatial reasoning tasks including measurement, perspective-taking and spatiotemporal tracking.

cs.CV↗

arXiv2025

Reasoning Path and Latent State Analysis for Multi-view Visual Spatial Reasoning: A Cognitive Science Perspective

Spatial reasoning is a core aspect of human intelligence that allows perception, inference and planning in 3D environments. However, current vision-language models (VLMs) struggle to maintain geometric coherence and cross-view consistency for spatial reasoning in multi-view settings. We attribute this gap to the lack of fine-grained benchmarks that isolate multi-view reasoning from single-view perception and temporal factors. To address this, we present ReMindView-Bench, a cognitively grounded benchmark for evaluating how VLMs construct, align and maintain spatial mental models across complementary viewpoints. ReMindView-Bench systematically varies viewpoint spatial pattern and query type to probe key factors of spatial cognition. Evaluations of 15 current VLMs reveals consistent failures in cross-view alignment and perspective-taking in multi-view spatial reasoning, motivating deeper analysis on the reasoning process. Explicit phase-wise analysis using LLM-as-a-judge and self-consistency prompting shows that VLMs perform well on in-frame perception but degrade sharply when integrating information across views. Implicit analysis, including linear probing and entropy dynamics, further show progressive loss of task-relevant information and uncertainty separation between correct and incorrect trajectories. These results provide a cognitively grounded diagnosis of VLM spatial reasoning and reveal how multi-view spatial mental models are formed, degraded and destabilized across reasoning phases. The ReMindView-Bench benchmark is available at https://huggingface.co/datasets/Xue0823/ReMindView-Bench, and the source codes of benchmark construction and VLM reasoning analysis are available at https://github.com/pittisl/ReMindView-Bench.

cs.AI↗

arXiv2025

Mass transfer and water management in proton exchange membrane fuel cells with a composite foam-rib flow field

Mass transfer capability of reactants and hydrothermal management is important for the performance and durability of proton exchange membrane fuel cells. In the conventional rib flow field, the oxygen transport is affected by the accumulation of under-rib liquid water which causes excessive concentration loss and limits cell performance. To improve the cell performance, a composite foam-rib flow field structure is proposed by combining the metal foam flow field and the conventional rib flow field. The proposed design is simulated by using a three-dimensional homogeneous non-isothermal numerical model. The results show that the composite foam-rib flow field, by improving the oxygen transfer and water removal capabilities under the ribs, can improve the oxygen concentration and current density without increasing the pumping power, thus improving the cell performance under different conditions. The key parameters of the composite foam-rib flow field are optimized. With the optimal metal foam filling ratio of 0.75 and porosity of 0.85, the peak power density and the limiting current density for the composite foam-rib flow field are higher than the conventional rib flow field by 5.20% and 22.68%.

physics.flu-dyn↗

arXiv2025

IE-Critic-R1: Advancing the Explanatory Measurement of Text-Driven Image Editing for Human Perception Alignment

Recent advances in text-driven image editing have been significant, yet the task of accurately evaluating these edited images continues to pose a considerable challenge. Different from the assessment of text-driven image generation, text-driven image editing is characterized by simultaneously conditioning on both text and a source image. The edited images often retain an intrinsic connection to the original image, which dynamically change with the semantics of the text. However, previous methods tend to solely focus on text-image alignment or have not well aligned with human perception. In this work, we introduce the Text-driven Image Editing Benchmark suite (IE-Bench) to enhance the assessment of text-driven edited images. IE-Bench includes a database contains diverse source images, various editing prompts and the corresponding edited results from different editing methods, and nearly 4,000 samples with corresponding Mean Opinion Scores (MOS) provided by 15 human subjects. Furthermore, we introduce IE-Critic-R1, which, benefiting from Reinforcement Learning from Verifiable Rewards (RLVR), provides more comprehensive and explainable quality assessment for text-driven image editing that aligns with human perception. Extensive experiments demonstrate IE-Critic-R1's superior subjective-alignments on the text-driven image editing task compared with previous metrics. Related data and codes are available to the public.

cs.CV↗

arXiv2025

Spatial Reasoning in Multimodal Large Language Models: A Survey of Tasks, Benchmarks and Methods

Spatial reasoning, which requires ability to perceive and manipulate spatial relationships in the 3D world, is a fundamental aspect of human intelligence, yet remains a persistent challenge for Multimodal large language models (MLLMs). While existing surveys often categorize recent progress based on input modality (e.g., text, image, video, or 3D), we argue that spatial ability is not solely determined by the input format. Instead, our survey introduces a taxonomy that organizes spatial intelligence from cognitive aspect and divides tasks in terms of reasoning complexity, linking them to several cognitive functions. We map existing benchmarks across text only, vision language, and embodied settings onto this taxonomy, and review evaluation metrics and methodologies for assessing spatial reasoning ability. This cognitive perspective enables more principled cross-task comparisons and reveals critical gaps between current model capabilities and human-like reasoning. In addition, we analyze methods for improving spatial ability, spanning both training-based and reasoning-based approaches. This dual perspective analysis clarifies their respective strengths, uncovers complementary mechanisms. By surveying tasks, benchmarks, and recent advances, we aim to provide new researchers with a comprehensive understanding of the field and actionable directions for future research.

cs.AI↗

arXiv2025

AIM: Software and Hardware Co-design for Architecture-level IR-drop Mitigation in High-performance PIM

SRAM Processing-in-Memory (PIM) has emerged as the most promising implementation for high-performance PIM, delivering superior computing density, energy efficiency, and computational precision. However, the pursuit of higher performance necessitates more complex circuit designs and increased operating frequencies, which exacerbate IR-drop issues. Severe IR-drop can significantly degrade chip performance and even threaten reliability. Conventional circuit-level IR-drop mitigation methods, such as back-end optimizations, are resource-intensive and often compromise power, performance, and area (PPA). To address these challenges, we propose AIM, comprehensive software and hardware co-design for architecture-level IR-drop mitigation in high-performance PIM. Initially, leveraging the bit-serial and in-situ dataflow processing properties of PIM, we introduce Rtog and HR, which establish a direct correlation between PIM workloads and IR-drop. Building on this foundation, we propose LHR and WDS, enabling extensive exploration of architecture-level IR-drop mitigation while maintaining computational accuracy through software optimization. Subsequently, we develop IR-Booster, a dynamic adjustment mechanism that integrates software-level HR information with hardware-based IR-drop monitoring to adapt the V-f pairs of the PIM macro, achieving enhanced energy efficiency and performance. Finally, we propose the HR-aware task mapping method, bridging software and hardware designs to achieve optimal improvement. Post-layout simulation results on a 7nm 256-TOPS PIM chip demonstrate that AIM achieves up to 69.2% IR-drop mitigation, resulting in 2.29x energy efficiency improvement and 1.152x speedup.

cs.AR↗

arXiv2025

Deciphering Personalization: Towards Fine-Grained Explainability in Natural Language for Personalized Image Generation Models

Image generation models are usually personalized in practical uses in order to better meet the individual users' heterogeneous needs, but most personalized models lack explainability about how they are being personalized. Such explainability can be provided via visual features in generated images, but is difficult for human users to understand. Explainability in natural language is a better choice, but the existing approaches to explainability in natural language are limited to be coarse-grained. They are unable to precisely identify the multiple aspects of personalization, as well as the varying levels of personalization in each aspect. To address such limitation, in this paper we present a new technique, namely \textbf{FineXL}, towards \textbf{Fine}-grained e\textbf{X}plainability in natural \textbf{L}anguage for personalized image generation models. FineXL can provide natural language descriptions about each distinct aspect of personalization, along with quantitative scores indicating the level of each aspect of personalization. Experiment results show that FineXL can improve the accuracy of explainability by 56\%, when different personalization scenarios are applied to multiple types of image generation models.

cs.LG↗

arXiv2025

Direct Observation and Optical Manipulation of Exciton-polariton Parametric Scattering Lasing in Temporal

The hybrid light-matter character of exciton-polaritons gives rise to distinct polariton parametric scattering (PPS) process, which holds promise for frontier applications in polaritonic quantum devices. However, the stable excitation and coherent optical manipulation of PPS remain challenging due to scattering bottlenecks and rapid dephasing effect in polariton many-body systems. In this study, we first report the direct observation and optical amplification of non-degenerate intermode PPS lasing at room temperature (RT). The specific polariton branch of strong-coupled nanobelt planar microcavity is resonantly excited by a near-infrared (NIR) femtosecond laser via two-photon absorption (TPA) scheme, and the non-degenerate signal- and idler-states are stimulated. Angle-resolved dispersion patterns clearly reveal the evolution of the pump-, signal-, and idler-states under different excitation powers. Based on our self-constructed ultrafast femtosecond resonant optical trigger set-up, a selective enhancement and modulation of the signal-state is realized. Furthermore, the dynamic measurements of nonlinear signal-state enhancement process demonstrate a sub-picosecond response time (0.4ps), confirming its potential for ultrafast optical manipulation. Our work establishes a platform for exploring TPA-driven PPS laser and provides a novel optical modulation route for polariton-based optoelectronic devices.

physics.optics↗

arXiv2025

Vector meson dominance in photon structure functions at small $x$ from holography

We investigate the photon structure functions via the photon-photon and photon-vector meson scattering within the framework of holographic QCD, focusing on the small Bjorken $x$ region and assuming that the Pomeron exchange dominates. The quasi-real photon structure functions are formulated as the convolution of the known U(1) vector field wave function with the Brower-Polchinski-Strassler-Tan (BPST) Pomeron exchange kernel in the five-dimensional AdS space. Assuming the vector meson dominance, the photon structure functions can be calculated in a different way with the BPST kernel and vector meson gravitational form factor, which can be obtained in a bottom-up AdS/QCD model, for the Pomeron-vector meson coupling. It is shown that the obtained $F_2$ structure functions in the both ways agree with the experimental data, which implies the realization of the vector meson dominance within the present model setup. Calculations for the longitudinal structure function and the longitudinal-to-transverse ratio are also presented.

hep-ph↗

arXiv2025

A Giant Peanut-shaped Ultra-High-Energy Gamma-Ray Emitter Off the Galactic Plane

Ultra-high-energy (UHE), exceeding 100 TeV (10^12 electronvolts), γ-rays manifests extreme particle acceleration in astrophysical sources. Recent observations by γ-ray telescopes, particularly by the Large High Altitude Air Shower Observatory (LHAASO), have revealed a few tens of UHE sources, indicating numerous Galactic sources capable of accelerating particles to PeV (10^15 electronvolts) energies. However, discerning the dominant acceleration mechanisms (leptonic versus hadronic), the relative contributions of specific source classes, and the role of particle transport in shaping their observed emission are central goals of modern UHE astrophysics. Here we report the discovery of a giant UHE γ-ray emitter at -17.5° off the Galactic plane - a region where UHE γ-ray sources are rarely found. The emitter exhibits a distinctive asymmetric shape, resembling a giant "Peanut" spanning 0.45° \times 4.6°, indicative of anisotropic particle distribution over a large area. A highly aged millisecond pulsar (MSP) J0218+4232 is the sole candidate accelerator positionally coincident with the Peanut region. Its association with UHE γ-rays extending to 0.7 PeV, if confirmed, would provide the first evidence of a millisecond pulsar powering PeV particles. Such a finding challenges prevailing models, which posit that millisecond pulsars cannot sustain acceleration to PeV energies. The detection reveals fundamental gaps in understanding particle acceleration, cosmic-ray transport, and interstellar magnetic field effects, potentially revealing new PeV accelerator (PeVatron) classes.

astro-ph.HE↗

arXiv2025

VGD: Visual Geometry Gaussian Splatting for Feed-Forward Surround-view Driving Reconstruction

Feed-forward surround-view autonomous driving scene reconstruction offers fast, generalizable inference ability, which faces the core challenge of ensuring generalization while elevating novel view quality. Due to the surround-view with minimal overlap regions, existing methods typically fail to ensure geometric consistency and reconstruction quality for novel views. To tackle this tension, we claim that geometric information must be learned explicitly, and the resulting features should be leveraged to guide the elevating of semantic quality in novel views. In this paper, we introduce \textbf{Visual Gaussian Driving (VGD)}, a novel feed-forward end-to-end learning framework designed to address this challenge. To achieve generalizable geometric estimation, we design a lightweight variant of the VGGT architecture to efficiently distill its geometric priors from the pre-trained VGGT to the geometry branch. Furthermore, we design a Gaussian Head that fuses multi-scale geometry tokens to predict Gaussian parameters for novel view rendering, which shares the same patch backbone as the geometry branch. Finally, we integrate multi-scale features from both geometry and Gaussian head branches to jointly supervise a semantic refinement model, optimizing rendering quality through feature-consistent learning. Experiments on nuScenes demonstrate that our approach significantly outperforms state-of-the-art methods in both objective metrics and subjective quality under various settings, which validates VGD's scalability and high-fidelity surround-view reconstruction.

cs.CV↗