This paper investigates the idea of encoding object-centered representations in the design of the reward function and policy architectures of a language-guided reinforcement learning agent. This is done using a combination of object-wise permutation invariant networks inspired from Deep Sets and gated-attention mechanisms. In a 2D procedurally-generated world where agents targeting goals in natural language navigate and interact with objects, we show that these architectures demonstrate strong generalization capacities to out-of-distribution goals. We study the generalization to varying numbers of objects at test time and further extend the object-centered architectures to goals involving relational reasoning.
@article{arxiv.2003.09443,
title = {Deep Sets for Generalization in RL},
author = {Tristan Karch and Cédric Colas and Laetitia Teodorescu and Clément Moulin-Frier and Pierre-Yves Oudeyer},
journal= {arXiv preprint arXiv:2003.09443},
year = {2020}
}
Comments
15 pages, 10 figures, published as a workshop Paper at ICLR: Beyond tabula rasa in RL (BeTR-RL). arXiv admin note: substantial text overlap with arXiv:2002.09253