Master student in SYSU

Education

• M.E., Computer Technology

  Sun Yat-sen University (Sept. 2021 - Present), Advisor: Xiaoxi Zhang

• B.E., Communication Engineering

  Sun Yat-sen University (Sept. 2017 - Jun. 2021)

Academic Experience

Smart Mobile Computing Lab, Sun Yat-sen University (Sept. 2021 - Present)

• Designed effective federated learning algorithms for dynamic edge networks (see publications)
• Implemented novel mechanisms into mainstream distributed/federated learning paradigms (working paper)
• Participated in projects on resource management for collaborative machine learning (supported by NSFC grants)

Academic Service

• Reviewer for IEEE TNSE 2023, BigComm 2022
• TA for Programming Design in Sun Yat-sen Univerisity 2022

Skills and Interests

• Programming: Python, MATLAB, Markdown, Latex
• Languages: Mandarin (native), Cantonese (native), English (IELTS: 7.5)
• Interests: Federated learning, optimization and algorithm design for machine learning

Projects

Leverage the Interplay of mini-batch size and aggregation frequency in Federated Learning

Publication

We propose a novel algorithm, named CoOptFL, by quantifying the interplay of mini-batch size and aggregation frequency to navigate the trade-offs among model convergence, completion time, and resource cost in federated learning. Based on which, we propose an online adaptive optimization algorithm AdaCoOpt by integrating the online estimates of the fluctuating edge network characteristics into CoOptFL. Experiments show a 37.6% reduction in the total training cost under a cost-sensitive scenario, and can achieve a 2.7%–7.9% higher final test accuracy than baselines

Publications

1. W. Liu, X. Zhang, J. Duan, C. Joe-Wong, Z. Zhou and X. Chen, DYNAMITE: Dynamic Interplay of Mini-Batch Size and Aggregation Frequency for Federated Learning with Static and Streaming Datasets IEEE Transactions on Mobile Computing.
2. W. Liu, X. Zhang, J. Duan, C. Joe-Wong, Z. Zhou and X. Chen, AdaCoOpt: Leverage the Interplay of Batch Size and Aggregation Frequency for Federated Learning, accepted to IEEE/ACM IWQoS 2023 (Best Paper Finalist).
3. W. Liu, X. Zhang, J. Duan, C. Joe-Wong, Z. Zhou and X. Chen, Federated Learning at the Edge: An Interplay of Mini-batch Size and Aggregation Frequency, accepted to IEEE INFOCOM 2023 FOGML WORKSHOP.

Honors and Awards

• IEEE/ACM International Symposium on Quality of Service 2023 Best Paper Finalist (2023)
• Excellent Postgraduate First Prize Scholarship × 2 (2021-2023)
• Excellent Undergraduate Second Prize Scholarship × 2 (2018-2020)
• Electronic Design Competition of Sun Yat-sen University (Third Prize) (2018)