About

I am an Associate Professor in the Department of Electrical and Computer Engineering of University of Virginia. I am also a member of the UVA Link Lab, and an active faculty in the Computer Engineering program. I got my B.S. and M.S. from Tsinghua University, and Ph.D. from UCLA. I received the NSF CAREER award in 2022.

Research topics

• Computer Science
• Computer network
• Artificial Intelligence
• Telecommunications
• Intensive care medicine
• Computer engineering
• Distributed computing
• Medicine
• Algorithm
• Pathology
• Virology

Selected publications

• Safety in Graph Machine Learning: Threats and Safeguards

  IEEE Transactions on Knowledge and Data Engineering · 2026-02-06

  article
  PublisherDOI

• A bandpass filtering method based on sparse representation for real-time ship detection

  Measurement · 2026-01-28

  article
  PublisherDOI

• MLorc: Momentum Low-rank Compression for Memory Efficient Large Language Model Adaptation

  arXiv (Cornell University) · 2025-06-02

  preprintOpen accessSenior author

  With increasing size of large language models (LLMs), full-parameter fine-tuning imposes substantial memory demands. To alleviate this, we propose a novel memory-efficient training paradigm called Momentum Low-rank compression (MLorc). The key idea of MLorc is to compress and reconstruct the momentum of matrix parameters during training to reduce memory consumption. Compared to LoRA, MLorc avoids enforcing a fixed-rank constraint on weight update matrices and thus enables full-parameter learning. Compared to GaLore, MLorc directly compress the momentum rather than gradients, thereby better preserving the training dynamics of full-parameter fine-tuning. We provide a theoretical guarantee for its convergence under mild assumptions. Empirically, MLorc consistently outperforms other memory-efficient training methods, matches or even exceeds the performance of full fine-tuning at small ranks (e.g., $r=4$), and generalizes well across different optimizers, all while not compromising time or memory efficiency.

  PublisherOA PDFDOI

• A Radical Heavy-Ball Method for Gradient Acceleration in Communication-Efficient Mobile Federated Learning

  IEEE Transactions on Mobile Computing · 2025-11-24 · 1 citations

  article

  Federated Learning (FL) is widely used in mobile computing as a communication-efficient distributed machine learning (ML) paradigm; however, it faces challenges such as model convergence to local optima or slow convergence due to the heterogeneity of client data. To mitigate data heterogeneity, the Nesterov Accelerated Gradient (NAG) method demonstrates its effectiveness by predictively updating the gradient to improve system performance. However, the performance of NAG depends heavily on the choice of decay coefficients; larger coefficients have greater acceleration but may lead to an unstable convergence process due to their unreasonable prediction of the descent gradient. To solve the above problems, this paper proposes the first radical heavy ball (RHB) method that combines momentum and NAG. In Stochastic Gradient Descent (SGD), momentum stabilizes the gradient descent process by integrating the historical gradients to update the parameters, and the RHB strategy decouples a single decay coefficient into an NAG component and a momentum component. The RHB introduces a gradient recall after each gradient acceleration by the NAG to strengthen the NAG's perception of the historical gradients, thus stabilizing the gradient descent process. By weighing the historical gradients and the predicted gradient, RHB effectively mitigates the instability of NAG convergence and demonstrates better performance. As a result, the algorithm further mitigates the impact of customer data heterogeneity in FL and can effectively deliver global update information to participants without additional communication costs. We conduct comprehensive experiments in a binary function, single node, and federated model environment to analyze the convergence properties in non-convex loss functions. RHB exhibits better performance and less computational overhead than many existing algorithms.

  PublisherDOI

• The 11th Mining and Learning from Time Series (MILETS): From Classical Methods to LLMs

  2025-08-03

  articleOpen access

  Time series data is now pervasive across domains such as healthcare, finance, entertainment, and transportation, driven by advances in sensing technologies that enable continuous data collection. The resulting increase in data volume and complexity poses significant challenges to traditional analysis methods, calling for the development of advanced, interdisciplinary approaches to temporal data mining. This workshop aims to: (1) identify key challenges in learning from time series data, including irregular sampling, spatiotemporal dependencies, and uncertainty quantification; (2) explore recent advances in algorithmic, statistical, theoretical, and systems-based solutions-ranging from classical methods to emerging techniques involving large language models (LLMs); and (3) foster collaboration by highlighting open problems and novel research directions in time series analysis. Bridging theory and practice, the workshop provides a platform for researchers and practitioners from academia, industry, and government to exchange ideas, discuss technical challenges, and showcase practical applications. Contributions from related areas such as AI, machine learning, data science, and statistics are strongly encouraged.

  PublisherOA PDFDOI

• Neighborhood and Global Perturbations Supported Sharpness-Aware Minimization in Federated Learning: From Local Tweaks to Global Awareness

  IEEE Transactions on Network Science and Engineering · 2025-12-01 · 1 citations

  article

  <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Federated Learning</b> (FL) can be coordinated under the orchestration of a central server to build a privacy-preserving model without collaborative data exchange. However, participant data heterogeneity leads to local optima divergence, affecting convergence outcomes. Recent research focused on global <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">sharpness-aware minimization</b> (SAM) and dynamic regularization to enhance consistency between global and local generalization and optimization objectives in FL. Nonetheless, the estimation of global SAM introduces additional computational and memory overhead. At the same time, the local dynamic regularizer cannot capture the global update state due to training isolation. This paper proposes a novel FL algorithm, FedTOGA, designed to consider optimization and generalization objectives while maintaining minimal uplink communication overhead. By linking local perturbations to global updates, we improve global generalization consistency. Additionally, by linking the dynamic regularizer to global updates, FedTOGA improves global gradient perception and strengthens optimization consistency. Crucially, global updates are directly delivered to clients, allowing them to incorporate global knowledge without communication and computational cost. We also propose neighborhood perturbation to enhance local perturbation, analyzing its strengths and working principles. Theoretical analysis shows FedTOGA achieves faster convergence <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$O(1/T)$</tex-math></inline-formula> on the non-convex function. Empirical studies demonstrate that FedTOGA outperforms existing algorithms, with a 1% accuracy increase and 30% faster convergence, achieving <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">SOTA</b>.

  PublisherDOI

• Graph Prompting for Graph Learning Models: Recent Advances and Future Directions

  2025-08-03

  articleOpen access

  Graph learning models have demonstrated great prowess in learning expressive representations from large-scale graph data in a wide variety of real-world scenarios. As a prevalent strategy for training powerful graph learning models, the ''pre-training, adaptation'' scheme first pre-trains graph learning models on unlabeled graph data in a self-supervised manner and then adapts them to specific downstream tasks. During the adaptation phase, graph prompting emerges as a promising approach that learns trainable prompts while keeping the pre-trained graph learning models unchanged. In this paper, we present a systematic review of recent advancements in graph prompting. First, we introduce representative graph pre-training methods that serve as the foundation step of graph prompting. Next, we review mainstream techniques in graph prompting and elaborate on how they design learnable prompts for graph prompting. Furthermore, we summarize the real-world applications of graph prompting from different domains. Finally, we discuss several open challenges in existing studies with promising future directions in this field.

  PublisherDOI

• SpaMCI-DL: A Hybrid Deep Learning Framework for Integrated Identification of Domains and Spatially Variable Genes in Spatial Transcriptomics

  2025-12-15

  article

  Spatial transcriptomics technologies enable the generation of gene expression profiles while retaining spatial coordinates. Identifying spatial domains and spatially variable genes (SVGs) are crucial tasks in spatial transcriptomics, offering valuable insights into biological functions. However, a deep learning framework that integrates SVGs detection with spatial domain identification is still lacking. In this study, we propose a multi-task ensemble analysis framework for spatial transcriptomics, named SpaMCI-DL, which adopts multi-constrained interpretable deep learning to jointly perform SVGs detection and spatial domain identification. SpaMCI-DL first employs a graph convolutional autoencoder to identify spatial domains by incorporating binary and graph structural constraints. Subsequently, based on the learned spatial domains, SpaMCI-DL utilizes a gradients-based method with multi-scale constraints to detect SVGs, enhancing the interpretability and biological relevance of the results. Comparative evaluations against state-of-the-art methods on five spatial transcriptomics datasets, spanning diverse species and tissues, demonstrate that SpaMCI-DL achieves superior performance in both spatial domain identification and SVGs detection. The code are available at https://github.com/liangxiao-cs/SpaMCI-DL.

  PublisherDOI

• Federated Split Learning With Improved Communication and Storage Efficiency

  IEEE Transactions on Mobile Computing · 2025-07-23 · 2 citations

  articleSenior author

  Federated learning (FL) is one of the popular distributed machine learning (ML) solutions but incurs significant communication and computation costs at edge devices. Federated split learning (FSL) can train sub-models in parallel and reduce the computational burden of edge devices by splitting the model architecture. However, it still requires a high communication overhead due to transmitting the smashed data and gradients between clients and the server in every global round. Furthermore, the server must maintain separate partial models for every client, leading to a significant storage requirement. To address these challenges, this paper proposes a novel communication and storage efficient federated split learning method, termed CSE-FSL, which utilizes an auxiliary network to locally update the weights of the clients while keeping a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">single</i> model at the server, hence avoiding frequent transmissions of gradients from the server and greatly reducing the storage requirement of the server. Additionally, a new model update method of transmitting the smashed data in selected epochs can reduce the amount of smashed data sent from the clients. We provide a theoretical analysis of CSE-FSL, rigorously guaranteeing its convergence under non-convex loss functions. The extensive experimental results further indicate that CSE-FSL achieves a significant communication reduction over existing FSL solutions using real-world FL tasks.

  PublisherDOI

• Chain-of-Thought Enhanced Shallow Transformers for Wireless Symbol Detection

  ArXiv.org · 2025-06-26

  preprintOpen accessSenior author

  Transformers have shown potential in solving wireless communication problems, particularly via in-context learning (ICL), where models adapt to new tasks through prompts without requiring model updates. However, prior ICL-based Transformer models rely on deep architectures with many layers to achieve satisfactory performance, resulting in substantial storage and computational costs. In this work, we propose CHain Of thOught Symbol dEtection (CHOOSE), a CoT-enhanced shallow Transformer framework for wireless symbol detection. By introducing autoregressive latent reasoning steps within the hidden space, CHOOSE significantly improves the reasoning capacity of shallow models (1-2 layers) without increasing model depth. This design enables lightweight Transformers to achieve detection performance comparable to much deeper models, making them well-suited for deployment on resource-constrained mobile devices. Experimental results demonstrate that our approach outperforms conventional shallow Transformers and achieves performance comparable to that of deep Transformers, while maintaining storage and computational efficiency. This represents a promising direction for implementing Transformer-based algorithms in wireless receivers with limited computational resources.

  PublisherOA PDFDOI

Recent grants

• CAREER: Towards a Communication Foundation for Distributed and Decentralized Machine Learning

  NSF · $500k · 2022–2027

• Collaborative Research: SWIFT: SMALL: Learning-Efficient Spectrum Access for No-Sensing Devices in Shared Spectrum

  NSF · $270k · 2020–2025

• CCSS: Collaborative Research: Towards a Resource Rationing Framework for Wireless Federated Learning

  NSF · $200k · 2020–2025

• Collaborative Research: MLWiNS: Dino-RL: A Domain Knowledge Enriched Reinforcement Learning Framework for Wireless Network Optimization

  NSF · $273k · 2020–2025

Frequent coauthors

• Jing Yang

  Zhejiang University of Technology

  34 shared

• Chengshuai Shi

  27 shared

• Mihaela van der Schaar

  25 shared

• Michael P. Fitz

  23 shared

• Jia Li

  21 shared

• Jie Xu

  Shandong University of Science and Technology

  19 shared

• Ruida Zhou

  17 shared

• Shaoting Ren

  14 shared

Labs

• Link LabPI

Education

• B.S.

  Tsinghua University

• M.S.

  Tsinghua University

• Ph.D.

  UCLA

Awards & honors

• NSF CAREER award