Offline Imitation Learning

Conservative bias correction

Bias correction approach that prioritizes safety by penalizing under-represented actions in the demonstration data.

← Indietro