Ph.D. student at McGill University and Mila - Quebec AI Institute.
About Me
I am a Ph.D. student in ECE at McGill University and Mila‑Quebec AI Institute, supervised by Mark Coates.
During my Ph.D., I was a visiting PhD student at Torr Vision Group, University of Oxford, working with Philip Torr; and have been interning in Huawei Noah’s Ark Lab Montréal.
Prior to that, I received my M.Sc. in CS from McGill and Mila, supervised by Reihaneh Rabbany and Adriana Romero Soriano; my B.Sc. in Stat from Sun Yat-sen University.
My primary research focus is on Geometric Deep Learning, particularly learning on graphs (on permutation group). I develop Graph Neural Networks (GNNs) that are both theoretically and empirically expressive, going beyond Message-passing Networks (MPNNs). This includes Graph Transformers and Continuous Convolution Kernels for learning graphs.
I am also interested in time-series forecasting and, more recently, have been exploring RL post-training for LLMs.
My research focus has shifted away from my earlier work in data mining.
Taken in Stamford Bridge
Selected Publication
My work has appeared in top-tier conferences (e.g., ICML, AISTATS, AAAI, KDD…), with 900+ citations.
- Advancing Multi-Agent RAG system with Minimalist Reinforcement Learning
Yihong Wu*, Liheng Ma*, Muzhi Li*, Jiaming Zhou, Lei Ding, Jianye HAO, Ho-fung Leung, Irwin King, Yingxue Zhang, Jian-Yun Nie
In Proc. Int. Conf. Auton. Agents Multi-Agent Syst. (AAMAS), 2026
- **It Takes Two: Your GRPO Is Secretly DPO**
Yihong Wu*, Liheng Ma*, Lei Ding, Muzhi Li, Xinyu Wang, Kejia Chen, Zhan Su, Zhanguang Zhang, Chenyang Huang, Yingxue Zhang, Mark Coates, Jian-Yun Nie
In Submission, 2025
- Plain Transformers Can be Powerful Graph Learners
Liheng Ma, Soumyasundar Pal, Yingxue Zhang, Philip H.S. Torr, Mark Coates
In Submission, 2025
- **SKOLR: Structured Koopman Operator Linear RNN for Time‑Series Forecasting**
Yitian Zhang, Liheng Ma, Antonios Valkanas, Boris N. Oreshkin, Mark Coates
In Proc. Int. Conf. Mach. Learn. (ICML), 2025
- **Multi‑Resolution Time‑Series Transformer for Long‑term Forecasting**
Yitian Zhang*, Liheng Ma*, Soumyasundar Pal, Yingxue Zhang, Mark Coates
In Proc. Int. Conf. Artif. Intell. Stat. (AISTATS), 2024
- CKGConv: General Graph Convolution with Continuous Kernels
Liheng Ma, Soumyasundar Pal, Yitian Zhang, Jiaming Zhou, Yingxue Zhang, Mark Coates
In Proc. Int. Conf. Mach. Learn. (ICML), 2024
- Graph Inductive Biases in Transformers without Message Passing
Liheng Ma*, Chen Lin*, Derek Lim, Adriana Romero‑Soriano, Puneet K. Dokania, Mark Coates, Philip H.S. Torr, Ser‑Nam Lim
In Proc. Int. Conf. Mach. Learn. (ICML), 2023
- **Memory Augmented Graph Neural Networks for Sequential Recommendation**
Chen Ma, Liheng Ma, Yingxue Zhang, Jianing Sun, Xue Liu, Mark Coates
In Proc. AAAI Conf. Artif. Intell. (AAAI), 2020
…
Full publication list can be found in my Google Scholar.
Academic Service
Conference Reviewer