Refactoring Policy for Compositional Generalizability using Self-Supervised Object Proposals

Part of Advances in Neural Information Processing Systems 33 (NeurIPS 2020)

AuthorFeedback Bibtex MetaReview Paper Review Supplemental


Tongzhou Mu, Jiayuan Gu, Zhiwei Jia, Hao Tang, Hao Su


We study how to learn a policy with compositional generalizability. We propose a two-stage framework, which refactorizes a high-reward teacher policy into a generalizable student policy with strong inductive bias. Particularly, we implement an object-centric GNN-based student policy, whose input objects are learned from images through self-supervised learning. Empirically, we evaluate our approach on four difficult tasks that require compositional generalizability, and achieve superior performance compared to baselines.