Learning Action Representations for Reinforcement Learning
arXiv:1902.00183
Abstract
Most model-free reinforcement learning methods leverage state representations (embeddings) for generalization, but either ignore structure in the space of actions or assume the structure is provided a priori. We show how a policy can be decomposed into a component that acts in a low-dimensional space of action representations and a component that transforms these representations into actual actions. These representations improve generalization over large, finite action sets by allowing the agent to infer the outcomes of actions similar to actions already taken. We provide an algorithm to both learn and use action representations and provide conditions for its convergence. The efficacy of the proposed method is demonstrated on large-scale real-world problems.
In Proceedings of the 36th International Conference on Machine Learning (ICML 2019)
References in corpus (5)
- Distributed Representations of Words and Phrases and their Compositionality
- Label-Embedding for Image Classification
- A Deep Reinforcement Learning Framework for the Financial Portfolio Management Problem
- Loss is its own Reward: Self-Supervision for Reinforcement Learning
- Learning to Factor Policies and Action-Value Functions: Factored Action Space Representations for Deep Reinforcement learning