Hierarchical Kickstarting for Skill Transfer in Reinforcement Learning
This addresses the challenge of improving RL agent performance in complex, sparse-reward environments by leveraging skill-based inductive biases, though it is incremental as it builds on existing skill transfer concepts.
The paper tackled the problem of incorporating predefined skills into reinforcement learning agents for complex environments with large state-action spaces and sparse rewards, and found that their novel method, Hierarchical Kickstarting (HKS), outperformed all evaluated baselines on the SkillHack benchmark based on NetHack.
Practising and honing skills forms a fundamental component of how humans learn, yet artificial agents are rarely specifically trained to perform them. Instead, they are usually trained end-to-end, with the hope being that useful skills will be implicitly learned in order to maximise discounted return of some extrinsic reward function. In this paper, we investigate how skills can be incorporated into the training of reinforcement learning (RL) agents in complex environments with large state-action spaces and sparse rewards. To this end, we created SkillHack, a benchmark of tasks and associated skills based on the game of NetHack. We evaluate a number of baselines on this benchmark, as well as our own novel skill-based method Hierarchical Kickstarting (HKS), which is shown to outperform all other evaluated methods. Our experiments show that learning with a prior knowledge of useful skills can significantly improve the performance of agents on complex problems. We ultimately argue that utilising predefined skills provides a useful inductive bias for RL problems, especially those with large state-action spaces and sparse rewards.