Bayesian Nonparametrics for Offline Skill Discovery
This work addresses the need for automated skill discovery without prior knowledge in offline reinforcement learning, offering a nonparametric solution that is incremental over existing methods.
The paper tackles the problem of offline skill discovery in reinforcement learning, where the number of skills is typically a fixed hyperparameter, by proposing a Bayesian nonparametric method that dynamically adjusts the number of options, and it demonstrates outperformance over state-of-the-art algorithms across various environments.
Skills or low-level policies in reinforcement learning are temporally extended actions that can speed up learning and enable complex behaviours. Recent work in offline reinforcement learning and imitation learning has proposed several techniques for skill discovery from a set of expert trajectories. While these methods are promising, the number K of skills to discover is always a fixed hyperparameter, which requires either prior knowledge about the environment or an additional parameter search to tune it. We first propose a method for offline learning of options (a particular skill framework) exploiting advances in variational inference and continuous relaxations. We then highlight an unexplored connection between Bayesian nonparametrics and offline skill discovery, and show how to obtain a nonparametric version of our model. This version is tractable thanks to a carefully structured approximate posterior with a dynamically-changing number of options, removing the need to specify K. We also show how our nonparametric extension can be applied in other skill frameworks, and empirically demonstrate that our method can outperform state-of-the-art offline skill learning algorithms across a variety of environments. Our code is available at https://github.com/layer6ai-labs/BNPO .