Sparse online maximum entropy inverse reinforcement learning via proximal optimization and truncated gradient

作者：

Highlights：

• A novel sparse online maximum entropy inverse reinforcement learning method is proposed.

• FTPRL and Truncated Gradient are introduced to solve overfitting and sparsity problems.

• Experiment results show the efficiency of our method in rewards learning of various problems.

摘要

•A novel sparse online maximum entropy inverse reinforcement learning method is proposed.•FTPRL and Truncated Gradient are introduced to solve overfitting and sparsity problems.•Experiment results show the efficiency of our method in rewards learning of various problems.

论文关键词：Maximum entropy,Inverse reinforcement learning,Proximal optimization,Truncated gradient,Q-learning,Regret bound

论文评审过程：Received 14 October 2021, Revised 14 June 2022, Accepted 11 July 2022, Available online 16 July 2022, Version of Record 30 July 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2022.109443