I am a 3rd-year PhD student (funded by EPSRC) @Game AI Group, Queen Mary University of London, supervised by Dr. Diego Perez and Dr. Alexander Dockhorn. I am a research intern at Machine Learning group, Microsoft Research Asia, supervised by Jinyu Wang and Lei Song.

Education

Queen Mary University of London : Sep.2020~To present

I work on state abstraction for strategy game playing. My current focus is developing general state abstraction methods that empowers complex game playing to better efficience and performance.

Nanchang University : Sep.2016~Jun.2020

B.E Software Engineering. GPA: 3.0/4.0. Graduation project: Gradient-based Intrinsic Motivation for Reinforcement Learning

Experience

• Microsoft Research Asia - Jan. 2023 ~ Present

  Machine Learning Group

  • Offline Reinforcement Learning.

• Southern University of Science and Technology - Oct. 2019 ~ Sept.2020

  Research Assistant at MEDIAN Group

  • Multi-task learning with RL.

• WestLake University - Aug. 2019 ~ Sept.2019

  Visiting Research Student at Robotic AI Group

  • Reinforcement Learning Agent for robot arm grasping in unstructured environment.

• OPPO Corp. - Sept. 2018 ~ Dec. 2018

  R&D at algorithm group of the Voice Assistant Team

  • Working on Natural Language Understanding (NLU) for OPPO’s first generation voice assistant.

Publication

• Mildly Constrained Evaluation Policy for Offline Reinforcement Learning paper, code, Linjie Xu, Zhengyao Jiang, Jinyu Wang, Lei Song and Jiang Bian

• Towards Applicable State Abstractions: a Preview in Strategy Games (RL as Agency workshop, RLDM 2022) Linjie Xu, Diego Perez-Liebana and Alexander Dockhorn.

• Elastic Monte Carlo Tree Search with State Abstraction for Strategy Game Playing (To appear at IEEE Conference On Games 2022) Linjie Xu, Alexander Dockhorn, Jorge Hurtado-Grueso, Dominic Jeurissen and Diego Perez Liebana

• A Hierarchial Approach for MARLO Challenge IEEE COG 2018. Linjie Xu, Yihong Chen

  We use a straightforward approach to solve 3 collaborative tasks on Minecraft, and got the 3rd place on MARLO challenge held by Microsoft Research, Cambridge.

Research Topics

• State Abstraction: I am trying to find a way to measure similarity between MDP components.
• Reinforcement Learning: I am trying to find a way to stablize consistent self-motivation.

Personal Interests

• Sports: Badminton, basketball and football. I am learning swimming.
• Collaboration: You are welcomed to send me an email to arrange a chat for research ideas, careers etc.
• Books:
  • Born a Crime by Trevor Noah
  • Good Economics for Hard Times by Abhijit V. Banerjee and Esther Duflo
  • Becoming by Michelle Obama
  • Lone Survivors by Marcus Luttrell