Searcharxiv⌕ Search

arXiv subjects

Lifeng Wang

Publications and source records attributed to Lifeng Wang.

At least 19 recordsLinked to original sources
arXiv2026

Efficient hot electron generation via low-coherence lasers

Hot electrons generated in laser-produced plasmas are a central focus in inertial confinement fusion, laboratory astrophysics, and high-energy-density physics. These electrons originate from instabilities in nonlinear laser-plasma interactions, which are critically modulated by laser bandwidth. Here, we experimentally demonstrate enhanced generation of hot electrons by utilizing instantaneous low-coherence lasers with two bandwidths (0.2% and 0.6%) at intensities of 2-8x10^{14} W/cm^2 and energies up to 620 J. A significant enhancement of hot electron temperature and hard X-ray yield is observed with the broadband lasers compared to a conventional narrowband laser. The results show that the hot electron energy conversion efficiency of the 0.6% broadband laser is approximately 4 times higher than that of the narrowband laser, reaching a maximum value of 2.8%. These findings validate a moderate-bandwidth laser as an efficient hot electron source and support the generation of bright X-ray sources for advanced imaging in high-energy-density physics.

physics.plasm-ph↗
arXiv2026

Scalable GNN-Based Power Allocation for Rate-Splitting Cell-Free Massive MIMO Systems

Cell-free massive multiple-input multiple-output (CF-mMIMO) systems provide enhanced coverage and capacity for next-generation wireless networks. However, CF-mMIMO systems face significant challenges in downlink power allocation (PA) due to imperfect channel state information (CSI), severe multi-user interference (MUI), and high computational complexity. To address these issues, rate-splitting multiple access (RSMA) is adopted as a robust interference management strategy. Accordingly, this paper proposes an unsupervised and scalable graph neural network (GNN) framework for PA in rate-splitting CF-mMIMO (RS-CF-mMIMO) systems, relying exclusively on large-scale fading (LSF) coefficients without instantaneous CSI. To resolve the dimensionality mismatch in dynamic networks, we introduce a slice-based adaptive layer that projects variable-dimension features into a fixed latent space. This mechanism enables a unified model to generalize across diverse topologies without retraining. Within this architecture, the sum spectral efficiency (SE) is maximized under per-AP power constraints, assuming maximum-ratio precoding for common streams and regularized zero-forcing precoding for private streams. We also derive a weighted minimum mean-square error-alternating direction method of multipliers (WMMSE-ADMM) algorithm as a performance upper bound. Extensive simulations verify that the proposed GNN framework achieves near-optimal SE and outperforms unsupervised deep neural networks (DNNs) across diverse system sizes and pilot assignment schemes. Furthermore, the scalable variant maintains robust performance while reducing the trainable parameter count by over 57% relative to DNNs and decreasing inference latency by up to three orders of magnitude compared with WMMSE-ADMM.

eess.SP↗
arXiv2026

SIREN: Unified Multi-Granularity Semantic Interaction for Multi-Modal Lifelong User Interest Modeling

Industrial recommender systems increasingly leverage lifelong user behavior histories and rich multi-modal content to capture evolving user preferences. However, effectively integrating multi-modal features into lifelong interest modeling remains challenging due to the inherent misalignment between multi-modal and collaborative spaces. Existing paradigms typically rely on separate modeling of multi-modal sequence and behavior sequence, and late fusion to alleviate the modality gap, which results in coarse-grained multi-modal representation and limited integration. In this paper, we propose SIREN, a unified multi-granularity semantic interaction framework for multi-modal lifelong user interest modeling. In the General Search Unit stage, we introduce two alternative retrieval strategies: multi-modal similarity-based soft retrieval for retrieval effectiveness, and Semantic ID (SemID)-based hard retrieval for efficient industrial serving. For the Exact Search Unit stage, we explicitly incorporate target-aware relevance via coarse similarity buckets and fine-grained prefix-encoded SemIDs, enabling unified interaction with collaborative ID features within the target-conditioned transformer architecture. Extensive experiments on the offline dataset demonstrate that SIREN achieves a state-of-the-art GAUC. Online A/B tests further demonstrate consistent GMV gains across multiple production scenarios, including +2.28% in Weixin Moments, +3.87% in Weixin Official Accounts, and +1.61% in Weixin Channels. From July 2025, SIREN has been fully launched for full-traffic serving in Tencent's advertising platform.

cs.IR↗
arXiv2026

RankUp: Towards High-rank Representations for Large Scale Advertising Recommender Systems

The scaling laws for recommender systems have been increasingly validated, where MetaFormer-based architectures consistently benefit from increased model depth, hidden dimensionality, and user behavior sequence length. However, whether representation capacity scales proportionally with parameter growth remains unexplored. Prior studies on RankMixer reveal that the effective rank of token representations exhibits a damped oscillatory trajectory across layers, failing to increase consistently with depth and even degrading in deeper layers. Motivated by this observation, we propose RankUp, an architecture designed to mitigate representation collapse and enhance expressive capacity through randomized permutation splitting over sparse features, a multi-embedding paradigm, global token integration and crossed pretrained embedding tokens. RankUp has been fully deployed in large-scale production across Weixin Video Accounts, Official Accounts and Moments, yielding GMV improvements of 3.41%, 4.81% and 2.12%, respectively.

cs.IR↗
arXiv2026

Anti-Jamming Optimization for EM-Compliant Active RIS via Decoupling Architecture

Wireless communication systems are increasingly vulnerable to sophisticated jamming attacks with the rapid evolution of jamming technologies and advanced signal processing techniques. While traditional anti-jamming techniques offer limited performance gains, active reconfigurable intelligent surfaces (RISs) have emerged as a promising channel-domain solution for improving resilience against jamming. Nonetheless, existing studies often rely on simplified electromagnetic (EM) models that do not fully capture mutual coupling (MC) and impedance mismatches in RIS hardware. In this paper, we propose an EM-compliant active (EMC-Active) RIS model for anti-jamming systems, explicitly incorporating the EM and physical properties at active RIS, such as MC effects, channel correlation, and discrete phase. To evaluate the anti-jamming performance of the proposed EMC-Active RIS, we develop a low-complexity alternating optimization (AO) algorithm based on the decoupling architecture (DA) to maximize the ergodic achievable rate. By leveraging the DA to explicitly eliminate MC effects among REs, the original coupled system is transformed into a tractable and scalable uncoupled representation. Numerical results demonstrate that the DA-based AO algorithm can significantly reduce the modeling and optimization complexity and efficiently solve the problem in an alternating manner with substantially reduced iteration overhead.

cs.IT↗
arXiv2026

Scalable Dexterous Robot Learning with AR-based Remote Human-Robot Interactions

This paper focuses on the scalable robot learning for manipulation in the dexterous robot arm-hand systems, where the remote human-robot interactions via augmented reality (AR) are established to collect the expert demonstration data for improving efficiency. In such a system, we present a novel method to address the general manipulation task problem. Specifically, the proposed method consists of two phases: i) In the first phase for pretraining, the policy is created in a behavior cloning (BC) manner, through leveraging the learning data from our AR-based remote human-robot interaction system; ii) In the second phase, a contrastive learning empowered reinforcement learning (RL) method is developed to obtain more efficient and robust policy than the BC, and thus a projection head is designed to accelerate the learning progress. An event-driven augmented reward is adopted for enhancing the safety. To validate the proposed method, both the physics simulations via PyBullet and real-world experiments are carried out. The results demonstrate that compared to the baselines, our method not only significantly speeds up the training process, but also achieves much better performance in terms of the success rate for fulfilling the manipulation tasks. By conducting the ablation study, it is confirmed that the proposed RL with contrastive learning overcomes policy collapse. Supplementary demonstrations are available at https://cyberyyc.github.io/.

cs.LG↗
arXiv2025

An Efficient Embedding Based Ad Retrieval with GPU-Powered Feature Interaction

In large-scale advertising recommendation systems, retrieval serves as a critical component, aiming to efficiently select a subset of candidate ads relevant to user behaviors from a massive ad inventory for subsequent ranking and recommendation. The Embedding-Based Retrieval (EBR) methods modeled by the dual-tower network are widely used in the industry to maintain both retrieval efficiency and accuracy. However, the dual-tower model has significant limitations: the embeddings of users and ads interact only at the final inner product computation, resulting in insufficient feature interaction capabilities. Although DNN-based models with both user and ad as input features, allowing for early-stage interaction between these features, are introduced in the ranking stage to mitigate this issue, they are computationally infeasible for the retrieval stage. To bridge this gap, this paper proposes an efficient GPU-based feature interaction for the dual-tower network to significantly improve retrieval accuracy while substantially reducing computational costs. Specifically, we introduce a novel compressed inverted list designed for GPU acceleration, enabling efficient feature interaction computation at scale. To the best of our knowledge, this is the first framework in the industry to successfully implement Wide and Deep in a retrieval system. We apply this model to the real-world business scenarios in Tencent Advertising, and experimental results demonstrate that our method outperforms existing approaches in offline evaluation and has been successfully deployed to Tencent's advertising recommendation system, delivering significant online performance gains. This improvement not only validates the effectiveness of the proposed method, but also provides new practical guidance for optimizing large-scale ad retrieval systems.

cs.LG↗
arXiv2025

Investigation of Inverse Bremsstrahlung Heating Driven by Broadband Lasers

Broadband lasers have become a key strategy for mitigating laser plasma instabilities in inertial confinement fusion, yet their impact on collisional inverse bremsstrahlung (IB) heating remains unclear. Using one-dimensional collisional particle-in-cell simulations, we systematically examine the effect of bandwidth-induced temporal incoherence on IB absorption in Au plasmas. The simulations are first benchmarked against classical absorption theory, verifying that the implemented Coulomb collision model accurately reproduces the theoretical IB heating rate. A direct comparison of the electron temperature evolution in the broadband and monochromatic cases shows that, although spectral broadening introduces transient picosecond-scale oscillations in the heating rate driven by stochastic intensity fluctuations, the long-term averaged heating and net IB absorption remain essentially unchanged.

physics.plasm-ph↗
arXiv2025

Yanyun-3: Enabling Cross-Platform Strategy Game Operation with Vision-Language Models

Cross-platform strategy game automation remains a challenge due to diverse user interfaces and dynamic battlefield environments. Existing Vision--Language Models (VLMs) struggle with generalization across heterogeneous platforms and lack precision in interface understanding and action execution. We introduce Yanyun-3, a VLM-based agent that integrates Qwen2.5-VL for visual reasoning and UI-TARS for interface execution. We propose a novel data organization principle -- combination granularity -- to distinguish intra-sample fusion and inter-sample mixing of multimodal data (static images, multi-image sequences, and videos). The model is fine-tuned using QLoRA on a curated dataset across three strategy game platforms. The optimal strategy (M*V+S) achieves a 12.98x improvement in BLEU-4 score and a 63% reduction in inference time compared to full fusion. Yanyun-3 successfully executes core tasks (e.g., target selection, resource allocation) across platforms without platform-specific tuning. Our findings demonstrate that structured multimodal data organization significantly enhances VLM performance in embodied tasks. Yanyun-3 offers a generalizable framework for GUI automation, with broader implications for robotics and autonomous systems.

cs.AI↗
arXiv2025

Boosting Pathology Foundation Models via Few-shot Prompt-tuning for Rare Cancer Subtyping

Rare cancers comprise 20-25% of all malignancies but face major diagnostic challenges due to limited expert availability-especially in pediatric oncology, where they represent over 70% of cases. While pathology vision-language (VL) foundation models show promising zero-shot capabilities for common cancer subtyping, their clinical performance for rare cancers remains limited. Existing multi-instance learning (MIL) methods rely only on visual features, overlooking cross-modal knowledge and compromising interpretability critical for rare cancer diagnosis. To address this limitation, we propose PathPT, a novel framework that fully exploits the potential of vision-language pathology foundation models through spatially-aware visual aggregation and task-specific prompt tuning. Unlike conventional MIL, PathPT converts WSI-level supervision into fine-grained tile-level guidance by leveraging the zero-shot capabilities of VL models, thereby preserving localization on cancerous regions and enabling cross-modal reasoning through prompts aligned with histopathological semantics. We benchmark PathPT on eight rare cancer datasets(four adult and four pediatric) spanning 56 subtypes and 2,910 WSIs, as well as three common cancer datasets, evaluating four state-of-the-art VL models and four MIL frameworks under three few-shot settings. Results show that PathPT consistently delivers superior performance, achieving substantial gains in subtyping accuracy and cancerous region grounding ability. This work advances AI-assisted diagnosis for rare cancers, offering a scalable solution for improving subtyping accuracy in settings with limited access to specialized expertise.

cs.CV↗
arXiv2025

Large Foundation Model for Ads Recommendation

Online advertising relies on accurate recommendation models, with recent advances using pre-trained large-scale foundation models (LFMs) to capture users' general interests across multiple scenarios and tasks. However, existing methods have critical limitations: they extract and transfer only user representations (URs), ignoring valuable item representations (IRs) and user-item cross representations (CRs); and they simply use a UR as a feature in downstream applications, which fails to bridge upstream-downstream gaps and overlooks more transfer granularities. In this paper, we propose LFM4Ads, an All-Representation Multi-Granularity transfer framework for ads recommendation. It first comprehensively transfers URs, IRs, and CRs, i.e., all available representations in the pre-trained foundation model. To effectively utilize the CRs, it identifies the optimal extraction layer and aggregates them into transferable coarse-grained forms. Furthermore, we enhance the transferability via multi-granularity mechanisms: non-linear adapters for feature-level transfer, an Isomorphic Interaction Module for module-level transfer, and Standalone Retrieval for model-level transfer. LFM4Ads has been successfully deployed in Tencent's industrial-scale advertising platform, processing tens of billions of daily samples while maintaining terabyte-scale model parameters with billions of sparse embedding keys across approximately two thousand features. Since its production deployment in Q4 2024, LFM4Ads has achieved 10+ successful production launches across various advertising scenarios, including primary ones like Weixin Moments and Channels. These launches achieve an overall GMV lift of 2.45% across the entire platform, translating to estimated annual revenue increases in the hundreds of millions of dollars.

cs.LG↗
arXiv2024

Knowledge-enhanced Pretraining for Vision-language Pathology Foundation Model on Cancer Diagnosis

Vision-language foundation models have shown great promise in computational pathology but remain primarily data-driven, lacking explicit integration of medical knowledge. We introduce KEEP (KnowledgE-Enhanced Pathology), a foundation model that systematically incorporates disease knowledge into pretraining for cancer diagnosis. KEEP leverages a comprehensive disease knowledge graph encompassing 11,454 diseases and 139,143 attributes to reorganize millions of pathology image-text pairs into 143,000 semantically structured groups aligned with disease ontology hierarchies. This knowledge-enhanced pretraining aligns visual and textual representations within hierarchical semantic spaces, enabling deeper understanding of disease relationships and morphological patterns. Across 18 public benchmarks (over 14,000 whole-slide images) and 4 institutional rare cancer datasets (926 cases), KEEP consistently outperformed existing foundation models, showing substantial gains for rare subtypes. These results establish knowledge-enhanced vision-language modeling as a powerful paradigm for advancing computational pathology.

eess.IV↗
arXiv2024

Structure of weakly collisional shock waves of multicomponent plasmas inside hohlraums of indirect inertial confinement fusions

In laser-driven indirect inertial confinement fusion (ICF), a hohlraum--a cavity constructed from high-Z materials--serves the purpose of converting laser energy into thermal x-ray energy. This process involves the interaction of low-density ablated plasmas, which can give rise to weakly collisional shock waves characterized by a Knudsen number $K_n$ on the order of 1. The Knudsen number serves as a metric for assessing the relative importance of collisional interactions. Preliminary experimental investigations and computational simulations have demonstrated that the kinetic effects associated with weakly collisional shock waves significantly impact the efficiency of the implosion process. Therefore, a comprehensive understanding of the physics underlying weakly collisional shock waves is essential. This research aims to explore the formation and fundamental structural properties of weakly collisional shock waves within a hohlraum, as well as the phenomena of ion mixing and ion separation in multicomponent plasmas. Weakly collisional shocks occupy a transition regime between collisional shock waves ($K_n \ll 1$) and collisionless shock waves ($K_n \gg 1$), thereby exhibiting both kinetic effects and hydrodynamic behavior. These shock waves are primarily governed by an electrostatic field, which facilitates significant electrostatic sheath acceleration and ion reflection acceleration. The differentiation of ions occurs due to the varying charge-to-mass ratios of different ion species in the presence of electrostatic field, resulting in the separation of ion densities, velocities, temperatures and concentrations. The presence of weakly collisional shock waves within the hohlraum is expected to affect the transition of laser energy and the overall efficiency of the implosion process.

physics.plasm-ph↗
arXiv2024

Edge Intelligence in Satellite-Terrestrial Networks with Hybrid Quantum Computing

This paper exploits the potential of edge intelligence empowered satellite-terrestrial networks, where users' computation tasks are offloaded to the satellites or terrestrial base stations. The computation task offloading in such networks involves the edge cloud selection and bandwidth allocations for the access and backhaul links, which aims to minimize the energy consumption under the delay and satellites' energy constraints. To address it, an alternating direction method of multipliers (ADMM)-inspired algorithm is proposed to decompose the joint optimization problem into small-scale subproblems. Moreover, we develop a hybrid quantum double deep Q-learning (DDQN) approach to optimize the edge cloud selection. This novel deep reinforcement learning architecture enables that classical and quantum neural networks process information in parallel. Simulation results confirm the efficiency of the proposed algorithm, and indicate that duality gap is tiny and a larger reward can be generated from a few data points compared to the classical DDQN.

cs.DC↗
arXiv2024

Spatial-spectral Cell-free Networks: A Large-scale Case Study

This paper studies the large-scale cell-free networks where dense distributed access points (APs) serve many users. As a promising next-generation network architecture, cell-free networks enable ultra-reliable connections and minimal fading/blockage, which are much favorable to the millimeter wave and Terahertz transmissions. However, conventional beam management with large phased arrays in a cell is very time-consuming in the higher-frequencies, and could be worsened when deploying a large number of coordinated APs in the cell-free systems. To tackle this challenge, the spatial-spectral cell-free networks with the leaky-wave antennas are established by coupling the propagation angles with frequencies. The beam training overhead in this direction can be significantly reduced through exploiting such spatial-spectral coupling effects. In the considered large-scale spatial-spectral cell-free networks, a novel subchannel allocation solution at sub-terahertz bands is proposed by leveraging the relationship between cross-entropy method and mixture model. Since initial access and AP clustering play a key role in achieving scalable large-scale cell-free networks, a hierarchical AP clustering solution is proposed to make the joint initial access and cluster formation, which is adaptive and has no need to initialize the number of AP clusters. After AP clustering, a subchannel allocation solution is devised to manage the interference between AP clusters. Numerical results are presented to confirm the efficiency of the proposed solutions and indicate that besides subchannel allocation, AP clustering can also have a big impact on the large-scale cell-free network performance at sub-terahertz bands.

cs.NI↗
arXiv2024

Hierarchical Reinforcement Learning Empowered Task Offloading in V2I Networks

Edge computing plays an essential role in the vehicle-to-infrastructure (V2I) networks, where vehicles offload their intensive computation tasks to the road-side units for saving energy and reduce the latency. This paper designs the optimal task offloading policy to address the concerns involving processing delay, energy consumption and edge computing cost. Each computation task consisting of some interdependent sub-tasks is characterized as a directed acyclic graph (DAG). In such dynamic networks, a novel hierarchical Offloading scheme is proposed by leveraging deep reinforcement learning (DRL). The inter-dependencies among the DAGs of the computation tasks are extracted using a graph neural network with attention mechanism. A parameterized DRL algorithm is developed to deal with the hierarchical action space containing both discrete and continuous actions. Simulation results with a real-world car speed dataset demonstrate that the proposed scheme can effectively reduce the system overhead.

cs.NI↗
arXiv2024

BugBlitz-AI: An Intelligent QA Assistant

The evolution of software testing from manual to automated methods has significantly influenced quality assurance (QA) practices. However, challenges persist in post-execution phases, particularly in result analysis and reporting. Traditional post-execution validation phases require manual intervention for result analysis and report generation, leading to inefficiencies and potential development cycle delays. This paper introduces BugBlitz-AI, an AI-powered validation toolkit designed to enhance end-to-end test automation by automating result analysis and bug reporting processes. BugBlitz-AI leverages recent advancements in artificial intelligence to reduce the time-intensive tasks of manual result analysis and report generation, allowing QA teams to focus more on crucial aspects of product quality. By adopting BugBlitz-AI, organizations can advance automated testing practices and integrate AI into QA processes, ensuring higher product quality and faster time-to-market. The paper outlines BugBlitz-AI's architecture, discusses related work, details its quality enhancement strategies, and presents results demonstrating its effectiveness in real-world scenarios.

cs.SE↗