Proper Laplacian Representation Learning

arXiv:2310.10833v2 Announce Type: replace
Abstract: The ability to learn good representations of states is essential for solving large reinforcement learning problems, where exploration, generalization, and transfer are particularly challenging. The Laplacian representation is a promising approach to address these problems by inducing informative state encoding and intrinsic rewards for temporally-extended action discovery and reward shaping. To obtain the Laplacian representation one needs to compute the eigensystem of the graph Laplacian, which is often approximated through optimization objectives compatible with deep …

abstract arxiv cs.ai cs.lg discovery encoding exploration good intrinsic learn reinforcement reinforcement learning representation representation learning state transfer type