Chenxia Han (韩晨夏)

Email: cxhan at cse.cuhk.edu.hk

[Google Scholar][Github][Resume]

Biography

I received my M.Phil. degree from The Chinese University of Hong Kong, advised by Prof. Eric Lo, and was fortunate to collaborate with Prof. Yao Lu. Before my graduate studies, I gained valuable industry experience working with the research team at TuSimple, supervised by Dr. Naiyan Wang.

My research bridges machine learning and systems, with a focus on algorithm-system co-design to build practical systems. Specifically, I work on:

  1. Efficient training frameworks to reduce compute costs, memory footprint, and communication overhead;
  2. Accelerated model inference for video processing (e.g., top-k queries, event queries);
  3. GPU resource optimization to maximize hardware utilization (e.g., SM occupancy, VRAM bandwidth).

Selected Publications

Scalable Complex Event Processing on Video Streams
Chenxia Han, Chaokun Chang, Srijan Srivastava, Yao Lu, Eric Lo
SIGMOD, 2025
[Paper] [Code] [BibTex]

SGDRC — Software-Defined Dynamic Resource Control for Concurrent DNN Inference on NVIDIA GPUs
Yongkang Zhang, Haoxuan Yu, Chenxia Han, Cheng Wang, Baotong Lu, Yunzhe Li, Zhifeng Jiang, Yang Li, Xiaowen Chu, Huaicheng Li
PPoPP, 2025
[Paper] [Code] [BibTex]

Top-K Deep Video Analytics: A Probabilistic Approach
Ziliang Lai, Chenxia Han, Chris Liu, Pengfei Zhang, Eric Lo, Ben Kao
SIGMOD, 2021
[Paper] [Code] [BibTex]

SimpleDet: A Simple and Versatile Distributed Framework for Object Detection and Instance Recognition
Yuntao Chen, Chenxia Han, Yanghao Li, Zehao Huang, Yi Jiang, Naiyan Wang, Zhaoxiang Zhang
JMLR, 2019
[Paper] [Code] [BibTex]