2024-05-01
Our paper developing a new formulation and a learning method for discontinuous graphon games has been accepted by ICML 2024.
2024-04-15
My PhD application is over. Huge thanks to everyone who helped me!
I will join MIT LIDS & IDSS in Fall 2024.
Excited to start new collaborations and research!
2024-01-16
Our paper on federated reinforcement learning with heterogeneous agents has been accepted by ICLR 2024.
2023-12-31
I graduated from Columbia University!
I will start my PhD study in Fall 2024.
Feel free to reach out if you want to collaborate or chat in between.
I help machines efficiently learn how to make decisions.
My long-term research goal is to combine efficient optimization methods and tailored machine learning frameworks with strong theoretical guarantees to address societal challenges.
With this goal in mind, my current research interests are in decision-making (reinforcement learning & control), multi-agent machine learning (federated learning & game), and optimization.
Riemannian Trust Region Methods for SC1 Minimization Journal of Scientific Computing, 2024.
Graphon Mean Field Games with a Representative Player: Analysis and Learning Algorithm ICML, 2024.
Finite-Time Analysis of On-Policy Heterogeneous Federated Reinforcement Learning ICLR, 2024.
A Single Online Agent Can Efficiently Learn Mean Field Games ECAI, 2024.
Riemannian Adaptive Regularized Newton Methods with Hölder Continuous Hessians Submitted to Mathematics of Operations Research, 2023.
PhD Application Reflection, For Future Reference Some reflections on my PhD application process, hoping to help future applicants.
Pandoc Filters Useful pandoc filters for an academic workflow
Obsidian Terminal Theme An Obsidian theme that emulates a terminal
EDAV Garden A digital garden for Exploratory Data Analysis and Visualization
Review vs Stream A research report on music reviews and streams
Salt Library A multi-functional, interactive library system
Wox Theme Library A theme library for Wox launcher
Note Box The published version of my personal knowledge base [work in progress]