arXiv subjects

Wenxue Li

Publications and source records attributed to Wenxue Li.

18 recordsLinked to original sources

arXiv2026

TIGER: Taming Identity, Geometry, and Generative Priors for High-Quality Face Video Restoration

Face Video Restoration (FVR) aims to recover high-fidelity facial videos from degraded input while preserving identity and semantic consistency across frames. Existing methods often struggle to simultaneously address three key challenges: identity shift, viewpoint-entangled guidance, and perceptual realism. To tackle these issues, we propose TIGER, a structured tri-prior fusion framework that Tames Identity, Geometry, and gEnerative pRiors for high-quality FVR. Specifically, an Identity Prior is first established by injecting subject-discriminative embeddings into the latent space, effectively anchoring the subject's identity against severe degradations. Then, to provide temporally consistent structural guidance for dynamic videos, TIGER constructs a Geometry Prior by lifting 2D reference cues into a disentangled 3D parameter space, creating a geometric anchor through cross-source parameter fusion. Moreover, to achieve maximum efficiency without compromising realism, we harness the video generation model's Generative Prior through a one-step rectified flow. We further design a progressive three-stage training optimization strategy that refines structural fidelity, textural reconstruction, and distribution-level realism to ensure robust optimization. We also construct a large-scale FVR dataset to facilitate robust training and standardized evaluation. Extensive experiments demonstrate that TIGER achieves state-of-the-art performance in both identity fidelity and temporal stability, delivering a high-quality, efficient and identity-consistent FVR. Project page: https://yzhoulv.github.io/Tiger/.

cs.CV↗

arXiv2026

PixelWizard: Towards Efficient High-Fidelity Video Generation at Ultra-Large Spatial Resolution

High-resolution video generation faces a coupled bottleneck of optimization instability and prohibitive computational costs. The massive expansion of the token sequence not only biases optimization toward local textures at the expense of global coherence, leading to structural collapse, but also imposes prohibitive training costs and severe inference latency. To address this, we propose PixelWizard, a framework that hierarchically decouples global structure modeling from fine-grained detail synthesis. PixelWizard first establishes a compact spatiotemporal anchor to concentrate dense structural priors, which then guides fine-grained generation at high resolution. This mitigates the local optimization bias to ensure structural stability without compromising high-frequency details. Leveraging this structural stability, we introduce Noise-Span Aligned Shortcut Training to break the inference bottleneck. By explicitly modeling the step size, this mechanism allows the model to traverse the generation trajectory with large steps. Crucially, we incorporate Exponential Index-Biased Sampling and Adaptive Noise-Span Calibration to align optimization with the shifted noise schedules of high-resolution grids, ensuring robust few-step inference without incurring the heavy overhead of distillation. Extensive experiments demonstrate that PixelWizard achieves superior visual quality while accelerating the generative sampling of native 2K/4K videos by over 10x.

cs.CV↗

arXiv2026

Nonlinear atomic tunnelling boosted by bright squeezed vacuum

Nonlinear optical processes, mediated by multiphoton interactions rather than single-photon response, are routinely exploited to enable a range of light-based functionalities in devices and applications. Nonlinear effects are enhanced through higher intensity fields, which is a limiting strategy owing to potential radiation damage. An alternative strategy relies on the fluctuation redistribution typical of quantum light, but experimental demonstrations at the most fundamental level have been limited. Here we report experimental nonlinear tunnelling ionization of isolated atoms, a pivotal nonlinear process that drives high-harmonic generation and forms the basis of attosecond science, boosted by quantum light -- bright squeezed vacuum (BSV). A BSV light with an average pulse energy of 300 nJ achieves an effective intensity equivalent to that of a coherent light with 7.1 {\textmu}J, demonstrating a more than 20-fold quantum boost in nonlinear effect from BSV light. This boost is revealed by matching the peaks of the photoelectron momentum spectra produced by the BSV and coherent light using angular streaking. Furthermore, we demonstrate control of the effective intensity of the BSV by tuning the correlation function at fixed average pulse energy, establishing a robust method to tailor nonlinear processes via quantum statistics rather than classical intensity scaling. These findings may facilitate the development of quantum-controlled strong-field dynamics using tailored quantum light sources.

quant-ph↗

arXiv2026

Multi-stage Flow Scheduling for LLM Serving

Meeting stringent Time-To-First-Token (TTFT) requirements is crucial for LLM applications. To improve efficiency, modern LLM serving systems adopt disaggregated architectures with diverse parallelisms, introducing complex multi-stage workflows involving reusable KV-block retrieval, collective communication, and P2D transfer. Flows from dependent stages overlap within and across requests on shared bottleneck links, making TTFT highly susceptible to network contention and necessitating stage-aware scheduling. Unfortunately, most existing works schedule flows in a stage-agnostic manner, leading to uncoordinated contention that constitutes a primary cause of SLO violations. In this paper, we present MFS, a holistic multi-stage flow scheduling mechanism designed to maximize TTFT SLO attainment. At its core, MFS approximates the Least-Laxity-First (LLF) scheduling policy without requiring precise knowledge of a request's remaining slack. It achieves this through a Defer-and-Promote principle implemented through a Reverse Multi-Level Queue (RMLQ) structure. By dynamically promoting task precedence as effective laxity diminishes, MFS prioritizes flows with less laxity while preventing requests with loose SLOs from prematurely consuming network bandwidth. We implement MFS as a pluggable module integrated into vLLM, and evaluate it on a 8-server, 32-GPU testbed as well as through large-scale simulations. Our results demonstrate that MFS effectively outperforms state-of-the-art baselines, improving the TTFT SLO attainment by 1.2x--2.4x.

cs.NI↗

arXiv2026

Vehicle-Mounted Mid-Infrared Dual-Comb Spectroscopy for On-Road Trace Gas Detection

Advances in mid-infrared (MIR) dual-comb spectroscopy (DCS) have significantly enhanced molecular detection in recent years. The capability of DCS to precisely identify and quantify atmospheric trace gases makes it attractive for field applications across the environmental, agricultural, energy, and industrial sectors. In particular, there is a growing demand for mobile and continuous gas monitoring in outdoor environments where emission sources and sinks are often episodic and spatially heterogeneous. However, the practical field-deployment of DCS on mobile platforms under realistic field conditions has remained limited. This study demonstrates for the first time a vehicle-mounted MIR DCS system that enables continuous mobile atmospheric sampling across multiple outdoor sites and road environments. The system exhibited a stable signal-to-noise performance during on-road operation, including expressway driving at speeds up to 100 km/h. Furthermore, natural-gas leakage sources were successfully located and a two-dimensional methane concentration field was reconstructed around a controlled release source. In the future, the system can be integrated into more mobile platforms, such as unmanned aerial vehicles, enabling flexible trace gas detection over urban-scale regions.

physics.optics↗

arXiv2025

Analyzing Communication Predictability in LLM Training

Effective communication is essential in distributed training, with predictability being one of its most significant characteristics. However, existing studies primarily focus on exploiting predictability through online profiling for runtime optimization, without a systematic understanding of it. In this work, we aim to systematically formulate communication predictability in distributed training, particularly in Large Language Models (LLMs) that utilize hybrid parallelism. Our analysis focuses on both traffic patterns and communication overhead. Specifically, we investigate predictable traffic patterns in typical LLMs and evaluate how various factors influence GPU utilization and effective bandwidth (two critical variables affecting communication overhead). Furthermore, we develop an analytical formulation to estimate communication overhead in LLM training, which is validated with high accuracy against empirical data. Leveraging this formulation, we propose a configuration tuning tool, ConfigTuner, to optimize training performance. Compared to Megatron-LM, the training configurations optimized by ConfigTuner demonstrate up to a 1.36$\times$ increase in throughput. Compared to Alpa, ConfigTuner generates the same configuration suggestion while significantly reducing the search complexity.

cs.NI↗

arXiv2025

MixNet: A Runtime Reconfigurable Optical-Electrical Fabric for Distributed Mixture-of-Experts Training

Mixture-of-Expert (MoE) models outperform conventional models by selectively activating different subnets, named experts, on a per-token basis. This gated computation generates dynamic communications that cannot be determined beforehand, challenging the existing GPU interconnects that remain static during the distributed training process. In this paper, we advocate for a first-of-its-kind system, called MixNet, that unlocks topology reconfiguration during distributed MoE training. Towards this vision, we first perform a production measurement study and show that the MoE dynamic communication pattern has strong locality, alleviating the requirement of global reconfiguration. Based on this, we design and implement a regionally reconfigurable high-bandwidth domain on top of existing electrical interconnects using optical circuit switching (OCS), achieving scalability while maintaining rapid adaptability. We have built a fully functional MixNet prototype with commodity hardware and a customized collective communication runtime that trains state-of-the-art MoE models with in-training topology reconfiguration across 32 A100 GPUs. Large-scale packet-level simulations show that MixNet delivers comparable performance as the non-blocking fat-tree fabric while boosting the training cost efficiency (e.g., performance per dollar) of four representative MoE models by 1.2x-1.5x and 1.9x-2.3x at 100 Gbps and 400 Gbps link bandwidths, respectively.

cs.NI↗

arXiv2025

RationalVLA: A Rational Vision-Language-Action Model with Dual System

A fundamental requirement for real-world robotic deployment is the ability to understand and respond to natural language instructions. Existing language-conditioned manipulation tasks typically assume that instructions are perfectly aligned with the environment. This assumption limits robustness and generalization in realistic scenarios where instructions may be ambiguous, irrelevant, or infeasible. To address this problem, we introduce RAtional MAnipulation (RAMA), a new benchmark that challenges models with both unseen executable instructions and defective ones that should be rejected. In RAMA, we construct a dataset with over 14,000 samples, including diverse defective instructions spanning six dimensions: visual, physical, semantic, motion, safety, and out-of-context. We further propose the Rational Vision-Language-Action model (RationalVLA). It is a dual system for robotic arms that integrates the high-level vision-language model with the low-level manipulation policy by introducing learnable latent space embeddings. This design enables RationalVLA to reason over instructions, reject infeasible commands, and execute manipulation effectively. Experiments demonstrate that RationalVLA outperforms state-of-the-art baselines on RAMA by a 14.5% higher success rate and 0.94 average task length, while maintaining competitive performance on standard manipulation tasks. Real-world trials further validate its effectiveness and robustness in practical applications. Our project page is https://irpn-eai.github.io/RationalVLA.

cs.RO↗

arXiv2025

Swift: Rethinking RDMA Control Plane for Elastic Computing

Elastic computing enables dynamic scaling to meet workload demands, and Remote Direct Memory Access (RDMA) enhances this by providing high-throughput, low-latency network communication. However, integrating RDMA into elastic computing remains a challenge, particularly in control plane operations for RDMA connection setup. This paper revisits the assumptions of prior work on high-performance RDMA for elastic computing, and reveals that extreme microsecond-level control plane optimizations are often unnecessary. By challenging the conventional beliefs on the slowness of user-space RDMA control plane and the difficulty of user-space RDMA resource sharing, we uncover new design opportunities. Our key insight is that user-space RDMA connection setup can be significantly improved with caching, while RDMA resources can be efficiently shared among processes using fork. In light of this, we propose Swift, a simple yet effective solution that co-designs RDMA with a serverless framework to optimize performance for elastic computing. At its very core, Swift handles cold and warm serverless requests by swiftly initializing the RDMA control plane with cache-optimized libibverbs, and manages fork requests by leveraging the RDMA's fork capability. Implemented with OpenWhisk, Swift delivers 30.56-46.50% higher average throughput and 18.55-37.21% lower latency, at a cost of 6.5% control plane overhead, compared to prior solutions.

cs.NI↗

arXiv2024

Neighbor Does Matter: Density-Aware Contrastive Learning for Medical Semi-supervised Segmentation

In medical image analysis, multi-organ semi-supervised segmentation faces challenges such as insufficient labels and low contrast in soft tissues. To address these issues, existing studies typically employ semi-supervised segmentation techniques using pseudo-labeling and consistency regularization. However, these methods mainly rely on individual data samples for training, ignoring the rich neighborhood information present in the feature space. In this work, we argue that supervisory information can be directly extracted from the geometry of the feature space. Inspired by the density-based clustering hypothesis, we propose using feature density to locate sparse regions within feature clusters. Our goal is to increase intra-class compactness by addressing sparsity issues. To achieve this, we propose a Density-Aware Contrastive Learning (DACL) strategy, pushing anchored features in sparse regions towards cluster centers approximated by high-density positive samples, resulting in more compact clusters. Specifically, our method constructs density-aware neighbor graphs using labeled and unlabeled data samples to estimate feature density and locate sparse regions. We also combine label-guided co-training with density-guided geometric regularization to form complementary supervision for unlabeled data. Experiments on the Multi-Organ Segmentation Challenge dataset demonstrate that our proposed method outperforms state-of-the-art methods, highlighting its efficacy in medical image segmentation tasks.

cs.CV↗

arXiv2024

SAM2-UNet: Segment Anything 2 Makes Strong Encoder for Natural and Medical Image Segmentation

Image segmentation plays an important role in vision understanding. Recently, the emerging vision foundation models continuously achieved superior performance on various tasks. Following such success, in this paper, we prove that the Segment Anything Model 2 (SAM2) can be a strong encoder for U-shaped segmentation models. We propose a simple but effective framework, termed SAM2-UNet, for versatile image segmentation. Specifically, SAM2-UNet adopts the Hiera backbone of SAM2 as the encoder, while the decoder uses the classic U-shaped design. Additionally, adapters are inserted into the encoder to allow parameter-efficient fine-tuning. Preliminary experiments on various downstream tasks, such as camouflaged object detection, salient object detection, marine animal segmentation, mirror detection, and polyp segmentation, demonstrate that our SAM2-UNet can simply beat existing specialized state-of-the-art methods without bells and whistles. Project page: \url{https://github.com/WZH0120/SAM2-UNet}.

cs.CV↗

arXiv2024

TP-DRSeg: Improving Diabetic Retinopathy Lesion Segmentation with Explicit Text-Prompts Assisted SAM

Recent advances in large foundation models, such as the Segment Anything Model (SAM), have demonstrated considerable promise across various tasks. Despite their progress, these models still encounter challenges in specialized medical image analysis, especially in recognizing subtle inter-class differences in Diabetic Retinopathy (DR) lesion segmentation. In this paper, we propose a novel framework that customizes SAM for text-prompted DR lesion segmentation, termed TP-DRSeg. Our core idea involves exploiting language cues to inject medical prior knowledge into the vision-only segmentation network, thereby combining the advantages of different foundation models and enhancing the credibility of segmentation. Specifically, to unleash the potential of vision-language models in the recognition of medical concepts, we propose an explicit prior encoder that transfers implicit medical concepts into explicit prior knowledge, providing explainable clues to excavate low-level features associated with lesions. Furthermore, we design a prior-aligned injector to inject explicit priors into the segmentation process, which can facilitate knowledge sharing across multi-modality features and allow our framework to be trained in a parameter-efficient fashion. Experimental results demonstrate the superiority of our framework over other traditional models and foundation model variants.

cs.CV↗

arXiv2024

Generalizing to Unseen Domains in Diabetic Retinopathy with Disentangled Representations

Diabetic Retinopathy (DR), induced by diabetes, poses a significant risk of visual impairment. Accurate and effective grading of DR aids in the treatment of this condition. Yet existing models experience notable performance degradation on unseen domains due to domain shifts. Previous methods address this issue by simulating domain style through simple visual transformation and mitigating domain noise via learning robust representations. However, domain shifts encompass more than image styles. They overlook biases caused by implicit factors such as ethnicity, age, and diagnostic criteria. In our work, we propose a novel framework where representations of paired data from different domains are decoupled into semantic features and domain noise. The resulting augmented representation comprises original retinal semantics and domain noise from other domains, aiming to generate enhanced representations aligned with real-world clinical needs, incorporating rich information from diverse domains. Subsequently, to improve the robustness of the decoupled representations, class and domain prototypes are employed to interpolate the disentangled representations while data-aware weights are designed to focus on rare classes and domains. Finally, we devise a robust pixel-level semantic alignment loss to align retinal semantics decoupled from features, maintaining a balance between intra-class diversity and dense class features. Experimental results on multiple benchmarks demonstrate the effectiveness of our method on unseen domains. The code implementations are accessible on https://github.com/richard-peng-xia/DECO.

cs.CV↗

arXiv2024

Diffusion Model Driven Test-Time Image Adaptation for Robust Skin Lesion Classification

Deep learning-based diagnostic systems have demonstrated potential in skin disease diagnosis. However, their performance can easily degrade on test domains due to distribution shifts caused by input-level corruptions, such as imaging equipment variability, brightness changes, and image blur. This will reduce the reliability of model deployment in real-world scenarios. Most existing solutions focus on adapting the source model through retraining on different target domains. Although effective, this retraining process is sensitive to the amount of data and the hyperparameter configuration for optimization. In this paper, we propose a test-time image adaptation method to enhance the accuracy of the model on test data by simultaneously updating and predicting test images. We modify the target test images by projecting them back to the source domain using a diffusion model. Specifically, we design a structure guidance module that adds refinement operations through low-pass filtering during reverse sampling, regularizing the diffusion to preserve structural information. Additionally, we introduce a self-ensembling scheme automatically adjusts the reliance on adapted and unadapted inputs, enhancing adaptation robustness by rejecting inappropriate generative modeling results. To facilitate this study, we constructed the ISIC2019-C and Dermnet-C corruption robustness evaluation benchmarks. Extensive experiments on the proposed benchmarks demonstrate that our method makes the classifier more robust across various corruptions, architectures, and data regimes. Our datasets and code will be available at \url{https://github.com/minghu0830/Skin-TTA_Diffusion}.

eess.IV↗

arXiv2024

Semi- and Weakly-Supervised Learning for Mammogram Mass Segmentation with Limited Annotations

Accurate identification of breast masses is crucial in diagnosing breast cancer; however, it can be challenging due to their small size and being camouflaged in surrounding normal glands. Worse still, it is also expensive in clinical practice to obtain adequate pixel-wise annotations for training deep neural networks. To overcome these two difficulties with one stone, we propose a semi- and weakly-supervised learning framework for mass segmentation that utilizes limited strongly-labeled samples and sufficient weakly-labeled samples to achieve satisfactory performance. The framework consists of an auxiliary branch to exclude lesion-irrelevant background areas, a segmentation branch for final prediction, and a spatial prompting module to integrate the complementary information of the two branches. We further disentangle encoded obscure features into lesion-related and others to boost performance. Experiments on CBIS-DDSM and INbreast datasets demonstrate the effectiveness of our method.

cs.CV↗

arXiv2023

Gleam: An RDMA-accelerated Multicast Protocol for Datacenter Networks

RDMA has been widely adopted for high-speed datacenter networks. However, native RDMA merely supports one-to-one reliable connection, which mismatches various applications with group communication patterns (e.g., one-to-many). While there are some multicast enhancements to address it, they all fail to simultaneously achieve optimal multicast forwarding and fully unleash the distinguished RDMA capabilities. In this paper, we present Gleam, an RDMA-accelerated multicast protocol that simultaneously supports optimal multicast forwarding, efficient utilization of the prominent RDMA capabilities, and compatibility with the commodity RNICs. At its core, Gleam re-purposes the existing RDMA RC logic with careful switch coordination as an efficient multicast transport. Gleam performs the one-to-many connection maintenance and many-to-one feedback aggregation, based on an extended multicast forwarding table structure, to achieve integration between standard RC logic and in-fabric multicast. We implement a fully functional Gleam prototype. With extensive testbed experiments and simulations, we demonstrate Gleam's significant improvement in accelerating multicast communication of realistic applications. For instance, Gleam achieves 2.9X lower communication time of an HPC benchmark application and 2.7X higher data replication throughput.

cs.NI↗

arXiv2020

Intrinsic feature between malignant tumor cells and human normal leukocytes with statistical decision tree analysis via Raman spectroscopy

In this study, the combination of a developing data mining technique called statistical decision tree analysis method and Raman spectroscopy was proposed to differentiate human normal leukocytes from malignant tumor cells. Statistical results obtained indicate this method possesses an admirable performance of a mean classification accuracy of 94.43% on the one hand, base adenine and amide I are recognized as potential characterizations of main- and subintrinsic biological difference in between on the other hand. Moreover, these certain Raman bands reflecting intrinsic physiological differences can be directionally extracted from whole fingerprint spectra and then provide a fast and accurate manipulation for spectrum identification.

physics.optics↗

arXiv2020

Sequential Transmission Matrix Evaluation Via Spatiotemporal Transmitted Modes Decomposition

The transmission matrix (TM) is a representation to describe the light scattering process through a scattering medium. The degree of control elements in TM is correlated with the capacity of evaluating enormous equations with tremendous number of unknow parameters, which seriously restricts the efficiency and accuracy of TM, and thus further confines applications such as image recovery. To completely remove this restriction, we propose decomposing TM and sequentially acquiring the dimension reduced decompositions regarding to the time and space invariance nature of transmitted field behind scattering medium. This proposed approach would not only have the ability to achieve high dimension transmission matrix with fully controllable elements, but also brings optimized signal-to-noise ratio during evaluation processing, which provides researchers a way to reach maximal focusing efficiency or image reconstruction resolution.

physics.optics↗