There Is No Turning Back: A Self-Supervised Approach for Reversibility-Aware Reinforcement Learning

Abstract

We propose to learn to distinguish reversible from irreversible actions for better informed decision-making in Reinforcement Learning (RL). From theoretical considerations, we show that approximate reversibility can be learned through a simple surrogate task: ranking randomly sampled trajectory events in chronological order. Intuitively, pairs of events that are always observed in the same order are likely to be separated by an irreversible sequence of actions. Conveniently, learning the temporal order of events can be done in a fully self-supervised way, which we use to estimate the reversibility of actions from experience, without any priors. We propose two different strategies that incorporate reversibility in RL agents, one strategy for exploration (RAE) and one strategy for control (RAC). We demonstrate the potential of reversibility-aware agents in several environments, including the challenging Sokoban game. In synthetic tasks, we show that we can learn control policies that never fail and reduce to zero the side-effects of interactions, even without access to the reward function.

References in corpus (10)

• IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
• Agent57: Outperforming the Atari Human Benchmark
• Imagination-Augmented Agents for Deep Reinforcement Learning
• AI Safety Gridworlds
• Safe Exploration in Markov Decision Processes
• Neural Predictive Belief Representations
• Munchausen Reinforcement Learning
• Can Temporal Information Help with Contrastive Self-Supervised Learning?
• Time Reversal as Self-Supervision
• Skip-Clip: Self-Supervised Spatiotemporal Representation Learning by Future Clip Order Ranking