Yuda Song

Hi, I am first year PhD student at Machine Learning Department, Carnegie Mellon University . I am coadvise by Aarti Singh and Drew Bagnell. I am also closely working with Wen Sun. I finished my master degree in MLD, advised by Kris Kitani. I completed my undergraduate at UC San Diego with CS and Math double major and I was advised by Sicun Gao.

My email is yudas at andrew dot cmu dot edu.

Google Scholar  /  CV  /  Github  /  Twitter

profile photo

I am broadly interested in machine learning, especially reinforcement learning. I am current focusing on provably statistically and computationally efficient settings and algorithms in RL.

Provable Benefits of Representational Transfer in Reinforcement Learning
(alphabetical order) Alekh Agarwal, Yuda Song, Wen Sun, Kaiwen Wang, Mengdi Wang, Xuezhou Zhang
COLT, 2023.

We prove the benefit of representation learning on diverse source environments which allows efficient learning on the source environment with the learned representation under the low-rank MDPs setting.
The Virtues of Laziness in Model-based RL: A Unified Objective and Algorithms
Anirudh Vemula, Yuda Song, Aarti Singh, J. Andrew Bagnell, Sanjiban Choudhury
ICML, 2023.
Hybrid RL: Using Both Offline and Online Data Can Make RL Efficient
Yuda Song*, Yifei Zhou*, Ayush Sekhari, J. Andrew Bagnell, Akshay Krishnamurthy, Wen Sun
ICLR, 2023.

Combining online data and offline data can solve RL with both statistical and computation efficiency. Experiments on Montezuma's Revenge reveals that hybrid RL works much better than pure online RL and pure offline RL.
Representation Learning for General-sum Low-rank Markov Games
Chengzhuo Ni, Yuda Song, Xuezhou Zhang, Chi Jin, Mengdi Wang
ICLR, 2023.
Efficient Reinforcement Learning in Block MDPs: A Model-free Representation Learning Approach
Xuezhou Zhang, Yuda Song, Masatoshi Uehara, Mengdi Wang, Alekh Agarwal, Wen Sun
ICML, 2022.
[code] [Talk at RL theory seminars]

An efficient rich-observation RL algorithm that learns to decode from rich observations to latent states (via adversarial training), while balancing exploration and exploitation.
No-regret Model-Based Meta RL for Personalized Navigation
Yuda Song, Ye Yuan, Wen Sun, Kris Kitani
L4DC, 2022.
Transform2Act: Learning a Transform-and-Control Policy for Efficient Agent Design
Ye Yuan, Yuda Song, Zhengyi Luo, Wen Sun, Kris Kitani
ICLR, 2022. Oral presentation
[Project Page]
PC-MLP: Model-based Reinforcement Learning with Policy Cover Guided Exploration
Yuda Song, Wen Sun
ICML, 2021

A simple provably efficient model-based algorithm that achieves competitive performance in both dense reward continuous control tasks and sparse reward control tasks that require efficient exploration.
Provably Efficient Model-based Policy Adaptation
Yuda Song, Aditi Mavalankar, Wen Sun, Sicun Gao
ICML, 2020
[Project Page] [code]

We study Sim-to-Real/policy transfer/policy adaptation under a model-based framework resulting an algorithm that enjoyes strong theoretical guarantees and excellent empirical performance.
Teaching Assistant

  • UCSD CSE291: Topics in Search and Optimization (Winter 2020)
  • UCSD CSE154: Deep Learning (Fall 2019)
  • UCSD CSE150: Introduction to AI: Search and Reasoning (Winter 2019, Spring 2020)
  • UCSD CSE30: Computer Organization and Systems Programming (Spring 2019, Winter 2018)
  • UCSD CSE11: Introduction to CS & OOP (Fall 2018)

  • Service

  • Reviewer: ICML (2021-), NeurIPS (2021-), ICLR (2022-), AAAI (2021-2022).
  • Top Reviewer: NeurIPS 2022

  • Source code from here