Siyuan Xu

School of Electrical Engineering and Computer Science, The Pennsylvania State University Google Scholar Email: spx5032@psu.edu

About me

I am a PhD student at the Pennsylvania State University, advised by Prof. Minghui Zhu.
I received the B.S. in Electrical Engineering from Xi'an Jiaotong University in 2019.

In my research, I am always thinking about how to develop an agent with a generalist policy or one that can adapt its policy model to multi-task with multi-environment. My recent research interests mainly focus on reinforcement learning, optimization, and robot learning.

Representative / Recent Works

• Meta-Reinforcement Learning with Universal Policy Adaptation: Provable Near-Optimality under All-task Optimum Comparator
  Siyuan Xu, Minghui Zhu
  Conference on Neural Information Processing Systems (NeurIPS) 2024.
  

• Online Constrained Meta-Learning: Provable Guarantees for Generalization
  Siyuan Xu, Minghui Zhu
  Conference on Neural Information Processing Systems (NeurIPS) 2023. (Spotlight, Top 3.6%)
  

• Efficient Gradient Approximation Method for Constrained Bilevel Optimization
  Siyuan Xu, Minghui Zhu
  AAAI Conference on Artificial Intelligence (AAAI) 2023. (Oral)
  

• Meta Value Learning for Fast Policy-Centric Optimal Motion Planning
  Siyuan Xu, Minghui Zhu
  Robotics: Science and Systems (RSS) 2022.