Return-Based Contrastive Representation Learning for Reinforcement Learning
This work addresses sample efficiency issues for reinforcement learning practitioners, offering an incremental improvement by integrating return-based feedback into auxiliary tasks.
The paper tackles the problem of slow representation learning and low sample efficiency in deep reinforcement learning by proposing a novel auxiliary task that uses returns to discriminate state-action pairs, resulting in improved performance in low data regimes on Atari games and DeepMind Control suite tasks.
Recently, various auxiliary tasks have been proposed to accelerate representation learning and improve sample efficiency in deep reinforcement learning (RL). However, existing auxiliary tasks do not take the characteristics of RL problems into consideration and are unsupervised. By leveraging returns, the most important feedback signals in RL, we propose a novel auxiliary task that forces the learnt representations to discriminate state-action pairs with different returns. Our auxiliary loss is theoretically justified to learn representations that capture the structure of a new form of state-action abstraction, under which state-action pairs with similar return distributions are aggregated together. In low data regime, our algorithm outperforms strong baselines on complex tasks in Atari games and DeepMind Control suite, and achieves even better performance when combined with existing auxiliary tasks.