SkillS: Adaptive Skill Sequencing for Efficient Temporally-Extended Exploration

Giulia Vezzani, Dhruva Tirumala, Markus Wulfmeier, Dushyant Rao, Abbas Abdolmaleki, Ben Moran, Tuomas Haarnoja, Jan Humplik, Roland Hafner, Michael Neunert, Claudio Fantacci, Tim Hertweck, Thomas Lampe, Fereshteh Sadeghi, Nicolas Heess, Martin Riedmiller

Keywords: Reinforcement Learning, Control, Skills, Priors, Hierarchical Reinforcement Learning

Abstract:

The ability to effectively reuse prior knowledge is a key requirement when building general and flexible Reinforcement Learning (RL) agents.
Skill reuse is one of the most common approaches, but current methods have considerable limitations. For example, fine-tuning an existing policy frequently fails, as the policy can degrade rapidly early in training, particularly in sparse reward tasks. In a similar vein, distillation of expert behavior can lead to poor results when given sub-optimal experts.
We compare several common approaches for skill transfer on multiple domains and in several different transfer settings, including under changes in task and system dynamics. We identify how existing methods can fail and introduce an alternative approach which sidesteps some of these problems.
Our approach learns to sequence existing temporally-abstract skills for exploration but learns the final policy directly from the raw experience. This conceptual split enables rapid adaptation and thus efficient data collection but without constraining the final solution. Our approach significantly outperforms many classical methods across a suite of evaluation tasks and we use a broad set of ablations to highlight the importance of different components of our method.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Reinforcement Learning (eg, decision and control, planning, hierarchical RL, robotics)