https://bair.berkeley.edu/blog/2020/10/13/supervised-rl/
Reinforcement learning is supervised learning on optimized data
Ben Eysenbach and Aviral Kumar and Abhishek Gupta Oct 13, 2020
“This suggests that we can simply use inverse RL to relabel data in arbitrary multi-task settings: inverse RL provides a theoretically grounded mechanism for sharing experience across tasks.”