Exploration by Distributional Reinforcement Learning

Abstract

We propose a framework based on distributional reinforcement learning and recent attempts to combine Bayesian parameter updates with deep reinforcement learning. We show that our proposed framework conceptually unifies multiple previous methods in exploration. We also derive a practical algorithm that achieves efficient exploration on challenging control tasks.

IJCAI 2018

References in corpus (4)

• Noisy Networks for Exploration
• Deep Exploration via Randomized Value Functions
• Bootstrapped Thompson Sampling and Deep Exploration
• Bayesian Policy Gradients via Alpha Divergence Dropout Inference

Cited by in corpus (3)

• The Potential of the Return Distribution for Exploration in RL
• The Second Type of Uncertainty in Monte Carlo Tree Search
• Weighted Entropy Modification for Soft Actor-Critic