Hong Xu, Henry
 |
Office: Room 914, Ho Sin-Hang Engineering Building |
Research Interests
I work on systems and networking. I lead the NetX Lab. Welcome to visit us!
Our research is/was supported by funding from Research Grants Council (RGC) of Hong Kong, The Chinese University of Hong Kong, City University of Hong Kong, Huawei, Microsoft Research Asia, Azure, Amazon Web Services, etc.
Current Projects
ML systems: ML/AI is important, but infra and deployment support is often the bottleneck to real-world use
Compound and complex AI systems, orchestrate the workflows of compound AI systems with more than the LLM, and complex AI systems with more than a linear reasoning path
Large-scale ML, faster, cheaper, and easier training and inference for large-scale ML workloads
Simulation, accurate and efficient performance simulation of distributed training
Related pubs: BlockLLM, ICPP’24, EMNLP’23, ATC’23, EuroSys’23, CoNEXT’18, SoCC’18
Networking: what would the future network look like?
Programmable dataplane, how can we better use the SmartNICs/DPUs?
ML-empowered network management, how can ML help reduce the human effort needed in network management tasks?
Related pubs: Yala@ASPLOS’25, Meili, NAssim@SIGCOMM’22, DQN@SIGCOMM’22, INFOCOM’24, CoNEXT’22, ATC’20, MM’23, INFOCOM’23
Learning: we also study theory and algorithm on the side
Learning to Optimize (L2O), how can we leverage DL to help solve optimization programs?
Related pubs: CVPR’24, INFOCOM’24
Recent Publications (full list)
“Performance Prediction of On-NIC Network Functions with Multi-Resource Contention and Traffic Awareness”, ASPLOS 2025
“Accelerating Distributed MoE Training and Inference with Lina”, USENIX ATC 2023
“Lyra: Elastic Scheduling for Deep Learning Clusters”, ACM EuroSys 2023
“Software-defined Network Assimilation: Bridging the Last Mile Towards Centralized Network Configuration Management with NAssim”, ACM SIGCOMM 2022 (Best Paper Award)
“NetKernel: Making Network Stack Part of the Virtualized Infrastructure”, USENIX ATC 2020, ACM HotNets 2017 (Project website)
Prospective Students and Current Openings
[04/2024] I am looking for 1-2 PhD students starting in fall 2025. The early admissions exercise has begun and deadline of first-round intake is May 12, 2024 (second-round is July 25). See here for more details.
News
11/2024. Yala is now open source.
10/2024. Serving on the PC of CoNEXT’25. Please submit!
10/2024. Yala accepted to ASPLOS’25! Congrats to Shaofeng and the rest of the team!
08/2024. Serving as the publicity co-chair of SIGCOMM’25 in Coimbra, Portugal.
08/2024. Serving on the PC of ICLR’25.
06/2024. Arlo accepted to ICPP’24! Congrats to Xin and the rest of the team!
05/2024. Serving on the PC of NeurIPS’24. Please submit!
03/2024. Open source GoMathL2O for our CVPR’24 paper. Also contains implementations of many baseline L2O methods.
03/2024. Serving on the PC of NSDI’25. Please submit!
02/2024. GoMathL2O, a robust learning to optimize method, accepted to CVPR’24! Congrats to Qingyu and Wei!
12/2023. Serving on the PC of ICNP’24 and ICDCS’24. Please submit.
12/2023. Three papers accepted to INFOCOM’24! Congrats to Qingyu, Wenlong, Yicheng, and the rest of the team!
11/2023. Our work on delta compression in data backup accepted to ASPLOS’24! Congrats to Hongming and the team!
10/2023. Our work on adaptive gating in MoE accepted to EMNLP’23! Congrats to Jiamin and the team!
07/2023. Moby accepted to ACM MM’23! Congrats to Jingzong, Yikang, and Libin!
06/2023. Our work on learning for non-convex optimization has been accepted to ICML's SynS & ML Workshop!