Adversarially Guided Actor-Critic
This work provides a strong specific gain in sample efficiency and exploration for researchers and practitioners working on deep reinforcement learning in complex environments, particularly those with rare rewards.
This paper introduces Adversarially Guided Actor-Critic (AGAC), a novel deep reinforcement learning algorithm that addresses sample inefficiency in complex environments by adding an adversarial component. The adversary mimics the actor, while the actor learns to solve the task and differentiate itself from the adversary's predictions, leading to more exhaustive exploration. AGAC outperforms current state-of-the-art methods on hard-exploration and procedurally-generated tasks.
Despite definite success in deep reinforcement learning problems, actor-critic algorithms are still confronted with sample inefficiency in complex environments, particularly in tasks where efficient exploration is a bottleneck. These methods consider a policy (the actor) and a value function (the critic) whose respective losses are built using different motivations and approaches. This paper introduces a third protagonist: the adversary. While the adversary mimics the actor by minimizing the KL-divergence between their respective action distributions, the actor, in addition to learning to solve the task, tries to differentiate itself from the adversary predictions. This novel objective stimulates the actor to follow strategies that could not have been correctly predicted from previous trajectories, making its behavior innovative in tasks where the reward is extremely rare. Our experimental analysis shows that the resulting Adversarially Guided Actor-Critic (AGAC) algorithm leads to more exhaustive exploration. Notably, AGAC outperforms current state-of-the-art methods on a set of various hard-exploration and procedurally-generated tasks.