
The author proposes the DIGIC framework to identify causal features directly from demonstration data distribution for domain generalization in imitation learning.


coremsg

Domain-Generalizable-Imitation-Learning-by-Causal-Discovery


Domain Generalizable Imitation Learning by Causal Discovery


title_rewrite


デモンストレーションデータ分布から因果関係を直接特定し、ドメイン汎用の模倣学習を実現するための新しいフレームワークであるDIGICを導入します。


digic-domain-generalizable-imitation-learning-by-causal-discovery


DIGIC: Domain Generalizable Imitation Learning by Causal Discovery



Causal discovery is utilized in imitation learning to identify causal features for domain generalization, leading to the DIGIC framework.
