Diversity-Driven Extensible Hierarchical Reinforcement Learning
arXiv:1811.04324
Abstract
Hierarchical reinforcement learning (HRL) has recently shown promising advances on speeding up learning, improving the exploration, and discovering intertask transferable skills. Most recent works focus on HRL with two levels, i.e., a master policy manipulates subpolicies, which in turn manipulate primitive actions. However, HRL with multiple levels is usually needed in many real-world scenarios, whose ultimate goals are highly abstract, while their actions are very primitive. Therefore, in this paper, we propose a diversity-driven extensible HRL (DEHRL), where an extensible and scalable framework is built and learned levelwise to realize HRL with multiple levels. DEHRL follows a popular assumption: diverse subpolicies are useful, i.e., subpolicies are believed to be more useful if they are more diverse. However, existing implementations of this diversity assumption usually have their own drawbacks, which makes them inapplicable to HRL with multiple levels. Consequently, we further propose a novel diversity-driven solution to achieve this assumption in DEHRL. Experimental studies evaluate DEHRL with five baselines from four perspectives in two domains; the results show that DEHRL outperforms the state-of-the-art baselines in all four aspects.
arXiv comment8 pages, 8 figures, In Proceedings of the 33rd National Conference on Artificial Intelligence, AAAI 2019, Honolulu, Hawaii, USA, January 27, 2019. Jianyi Wang is the co-first author
References in corpus (11)
- Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
- Predicting Head Movement in Panoramic Video: A Deep Reinforcement Learning Approach
- Variational Intrinsic Control
- Learning and Transfer of Modulated Locomotor Controllers
- Strategic Attentive Writer for Learning Macro-Actions
- Latent Space Policies for Hierarchical Reinforcement Learning
- Multi-Level Discovery of Deep Options
- EX2: Exploration with Exemplar Models for Deep Reinforcement Learning
- Option Discovery in Hierarchical Reinforcement Learning using Spatio-Temporal Clustering
- Generalization Tower Network: A Novel Deep Neural Network Architecture for Multi-Task Learning
- Learning Approximate Stochastic Transition Models