Yingru LI

Ph.D. Candidate

The Chinese University of Hong Kong

Biography

Mr. Yingru Li is a Ph.D. Candidate in The Chinese University of Hong Kong (CUHK), Shenzhen, China. Fortunately, he is advised by Zhi-Quan (Tom) Luo. He received the bachelor degree in Computer Science (ACM Honors Program) from Huazhong University of Science and Technology with an advisory of Kun He. He was a research visiting student at Cornell University with John E. Hopcroft. His Ph.D. research is supported by SRIBD Ph.D. Fellowship, Presidential Ph.D. Fellowship and Tencent AI Ph.D. Fellowship.

He organized Reinforcement Learning Seminar in CUHK-SZ from 2019 to 2023.

Now actively seeking postdoctoral & research positions! my resumé.

July 2024: Will deliver an Invited Long Talk at the International Symposium on Mathematical Programming (ISMP), Montréal. The ISMP is the leading triennial conference focusing on mathematical optimization.
May 2024: AISTATS, Valencia, Spain. Our paper offers a groundbreaking prior-dependent analysis of PSRL under a linear mixture model. This helps understand how integrating prior knowledge like historical data or pre-trained models (LLMs) enhances RL agent efficiency.
May 2024: Showcasing HyperAgent at the ICLR in Vienna, Austria, during the Workshop on Bridging the Gap Between Practice and Theory in Deep Learning. HyperAgent represents a significant stride towards aligning theoretical foundations with practical deep RL applications.
March 2024: Two Talks at the Informs Optimization Society (IOS) Conference at Rice University. (1) “HyperAgent: A simple, efficient, scalable and provable RL framework for complex environments”.
Jan 2024: Our work about HyperAgent received Best Paper Award in the third doctoral and postdoctoral Daoyuan academic forum.
December 2023: NeurIPS, New Orleans 🚀 My research addresses efficiency challenges in reinforcement learning (RL). It encompasses both theoretical aspects of high-dimentional probability and practical applications in Deep RL [1]. I have developed a novel random projection tool for high-dimensional sequentially dependent data, a non-trivial martingale extension of Johnson–Lindenstrauss [2]. 🚀

Interests

Sequential Decision-making & Reinforcement Learning
Algorithms Design & Analysis
Mathematical Optimization
Applied probability
AI, Robotics & OR

Education

Ph.D. in Computer and Information Engineering, 2018 - present
The Chinese University of Hong Kong
B.Eng. in Computer Science (Honors Program). Outstanding Graduate, 2017
Huazhong University of Science and Technology, China

Featured Publications

Yingru Li, Jiawei Xu, Zhi-Quan Luo

January, 2024

Approximate Thompson sampling via Hypermodel and Index sampling

Hypermodel for efficient incremental approximation of the posterior (uncertainty quantification) over complex models without leveraging conjugacy as encountering more data; Index sampling for approximate posterior sampling for data-efficient sequential decision-making. This approach is provably superior than ensemble sampling and Langevin Monte-carlo.

Yingru Li, Jiawei Xu, Lei Han, Zhiquan Luo

January, 2024 The 41st International Conference on Machine Learning (ICML) (Submitted)

HyperAgent: A Simple, Scalable, Efficient and Provable Reinforcement Learning Framework for Complex Environments

Addressing data and computation efficiency challenge in RL for real-world problems with theoretical advancements and practical algorithm designs.(1) Theoretically, we demonstrate exponential improvements in per-step computation complexity; (2) Empirically, HyperAgent has significant practical efficiency gains, particularly in deep RL benchmarks.

Yingru Li

January, 2024

Probability Tools for Sequential Random Projection

First probabilistic framework for sequential random projection, an approach rooted in the challenges of sequential decision-making under uncertainty; A non-trivial martingale extension of Johnson-Lindenstrauss (JL) to sequentially adaptive data processes.

Recent Publications

Quickly discover relevant content by filtering publications.

Yingru Li, Liangqi Liu, Wenqiang Pu, Zhi-Quan Luo (2024). Optimistic Thompson Sampling for No-Regret Learning in Unknown Games. The 41st International Conference on Machine Learning (ICML) (Submitted).

Cite arXiv

Yingru Li, Zhi-Quan Luo (2024). Prior-dependent analysis of posterior sampling reinforcement learning with function approximation. The 27th International Conference on Artificial Intelligence and Statistics (AISTATS).

Cite

Yingru Li (2024). Simple, unified analysis of Johnson-Lindenstrauss with applications. The 37th Annual Conference on Learning Theory (COLT) (Submitted).

Cite arXiv

Yingru Li, Jiawei Xu, Zhiquan Luo (2023). Efficient and scalable reinforcement learning via hypermodel. NeurIPS 2023 Workshop on Adaptive Experimental Design and Active Learning in the Real World.

PDF Cite Poster Slides Website

Ziniu Li, Yingru Li, Yushun Zhang, Tong Zhang, Zhi-Quan Luo (2022). HyperDQN: A Randomized Exploration Method for Deep Reinforcement Learning. International Conference on Learning Representations (ICLR).

Cite ICLR2022

See all publications