VariBAD: A Very Good Method for Bayes-Adaptive Deep RL via Meta-Learning
This addresses the challenge of exploration-exploitation trade-offs in unknown environments for reinforcement learning practitioners, representing an incremental improvement over prior meta-RL methods.
The paper tackles the intractable problem of computing Bayes-optimal policies in reinforcement learning by introducing variBAD, a meta-learning method that approximates inference and incorporates task uncertainty into action selection. It demonstrates higher online return than existing methods in MuJoCo domains, though specific numerical gains are not provided.
Trading off exploration and exploitation in an unknown environment is key to maximising expected return during learning. A Bayes-optimal policy, which does so optimally, conditions its actions not only on the environment state but on the agent's uncertainty about the environment. Computing a Bayes-optimal policy is however intractable for all but the smallest tasks. In this paper, we introduce variational Bayes-Adaptive Deep RL (variBAD), a way to meta-learn to perform approximate inference in an unknown environment, and incorporate task uncertainty directly during action selection. In a grid-world domain, we illustrate how variBAD performs structured online exploration as a function of task uncertainty. We further evaluate variBAD on MuJoCo domains widely used in meta-RL and show that it achieves higher online return than existing methods.