Inverse Delayed Reinforcement Learning

This work introduces an inverse reinforcement learning framework designed to extract rewarding features from expert trajectories affected by delayed disturbances. It employs an efficient off-policy adversarial training scheme to derive expert features and recover optimal policies from augmented delayed observations, validated in MuJoCo environments under diverse delay settings.

Authors: Simon Sinong Zhan, Qingyuan Wu, Zhian Ruan, Frank Yang, Philip Wang, Yixuan Wang, Ruochen Jiao, Chao Huang, Qi Zhu (*equal contribution)

@article{zhan2024inverse, title={Inverse Delayed Reinforcement Learning}, author={Zhan, Simon Sinong and Wu, Qingyuan and Ruan, Zhian and Yang, Frank and Wang, Philip and Wang, Yixuan and Jiao, Ruochen and Huang, Chao and Zhu, Qi}, journal={arXiv preprint arXiv:2412.02931}, year={2024} }