Kun Lei

Master of Southwest Jiao-tong University

I am a research assistant at Shanghai Qizhi institute, where I am fortunate to work with Prof. Huazhe Xu at Tsinghua University. Before that, I was a research assistant at Westlake University, worked with Prof. Donglin Wang.

I received my master's degree from Southwest Jiao-tong University in June 2022, where I had the pleasue to work with Prof. Peng Guo at Southwest Jiao-tong University and Prof. Yi Wang at Auburn University Montgomery. I received my bachelor's degree from Chengdu University in June 2019.

Research interest

I am currently interested in research on the sample efficiency and adaptation capabilities of deep reinforcement learning algorithms and agents. My research objective is to create reliable, robust, and efficient algorithms to tackle a diverse range of sequential decision-making problems in real-world scenarios. I focus on reinforcement learning (especially offline RL) and robotics.

Previously, I conducted research on neural solvers for combinatorial optimization problems, i.e., in the field of AI for Operations Research. These neural agents have the potential to solve NP-hard problems with linear time complexity, see Zhihu.

Publications and preprints (Offline & online RL algorithms and robotics)

Papers sorted by recency, * indicates equal contribution.

Uni-O4: Unifying Online and Offline Deep Reinforcement Learning with Multi-Step On-Policy Optimization
Kun Lei, Zhengmao He*, Chenhao Lu*, Kaizhe Hu, Yang Gao, Huazhe Xu.
International Conference on Learning Representations (ICLR), 2024
with a nice rating 6688 / project page / arXiv / code / twitter
Behavior proximal policy optimization
Zifeng Zhuang*, Kun Lei*, Jinxin Liu, Donglin Wang, Yilang Guo.
International Conference on Learning Representations (ICLR), 2023
arXiv / code
Learning Visual Quadrupedal Loco-Manipulation from Demonstrations
Zhengmao He, Kun Lei, Yanjie Ze, Zhongyu Li, Huazhe Xu.
In submission
proj page / arXiv

Publications and preprints (AI for Operations Research)

Papers sorted by recency.

Large-scale Dynamic Scheduling for Flexible Job-shop with Random New Job Arrival by Hierarchical RL.
Kun Lei, Peng Guo, Yi Wang, Jian Zhang, Xiangyin Meng, Linmao Qian.
IEEE Transactions on Industrial Informatics (TII) , 2023
A Multi-action Deep Reinforcement Learning Framework for Flexible Job-shop Scheduling Problem.
Kun Lei, Peng Guo, Wenchao Zhao, Yi Wang, Linmao Qian, Xiangyin Meng.
Expert Systems with Applications (ESWA), 2022
paper / code / covered by 运筹OR帷幄智能制造与智能调度PaperWeekly

Misc. open-source projects

Behavior proximal policy optimization⭐ 54
Public code release for "BPPO", Feb 2023
End-to-end-DRL-for-FJSP.⭐ 168
Public code release for "A Multi-action Deep Reinforcement Learning Framework for Flexible Job-shop Scheduling Problem", Sep 2022
DRL-and-graph-neural-network-for-routing-problems.⭐ 112
Public code release for "Solve routing problems with a residual edge-graph attention neural network" June 2022
Dispatching-rules-for-FJSP.⭐ 54
Coed for dispatching rules for flexible job-shop scheduling problems", Sep 2022
FJSP-benchmarks.⭐ 27
The public benchmark instances of flexible job shop scheduling problem, Feb 2022
MIP-model-for-FJSP-and-solved-by-Gurobi.⭐ 17
Coed for solving MIP model using Gurobi", Feb 2022

I am from Si Chuan, China. My name in chinese is 雷坤. I used to live in the charming cities of Chengdu and Hangzhou in China. Currently, I live in the vibrant city of Shanghai (as depicted in the background picture).

======= >>>>>>> 1af91862845d734e3546ef6b98398f75f88fa701