Information-Directed Exploration for Deep Reinforcement Learning
This addresses exploration inefficiencies in reinforcement learning for AI agents, though it is an incremental advance building on existing methods.
The paper tackles the challenge of efficient exploration in deep reinforcement learning by proposing Information-Directed Sampling (IDS) to account for heteroscedastic noise, resulting in significant improvements over alternative approaches on Atari games.
Efficient exploration remains a major challenge for reinforcement learning. One reason is that the variability of the returns often depends on the current state and action, and is therefore heteroscedastic. Classical exploration strategies such as upper confidence bound algorithms and Thompson sampling fail to appropriately account for heteroscedasticity, even in the bandit setting. Motivated by recent findings that address this issue in bandits, we propose to use Information-Directed Sampling (IDS) for exploration in reinforcement learning. As our main contribution, we build on recent advances in distributional reinforcement learning and propose a novel, tractable approximation of IDS for deep Q-learning. The resulting exploration strategy explicitly accounts for both parametric uncertainty and heteroscedastic observation noise. We evaluate our method on Atari games and demonstrate a significant improvement over alternative approaches.