Learning to reinforcement learn
arXiv:1611.05763
Abstract
In recent years deep reinforcement learning (RL) systems have attained superhuman performance in a number of challenging task domains. However, a major limitation of such applications is their demand for massive amounts of training data. A critical present objective is thus to develop deep RL methods that can adapt rapidly to new tasks. In the present work we introduce a novel approach to this challenge, which we refer to as deep meta-reinforcement learning. Previous work has shown that recurrent networks can support meta-learning in a fully supervised context. We extend this approach to the RL setting. What emerges is a system that is trained using one RL algorithm, but whose recurrent dynamics implement a second, quite separate RL procedure. This second, learned RL algorithm can differ from the original one in arbitrary ways. Importantly, because it is learned, it is configured to exploit structure in the training domain. We unpack these points in a series of seven proof-of-concept experiments, each of which examines a key aspect of deep meta-RL. We consider prospects for extending and scaling up the approach, and also point out some potentially important implications for neuroscience.
17 pages, 7 figures, 1 table
References in corpus (2)
Cited by in corpus (9)
- Meta-SGD: Learning to Learn Quickly for Few-Shot Learning
- Searching for Activation Functions
- Structure Learning in Motor Control:A Deep Reinforcement Learning Model
- Building Machines that Learn and Think for Themselves: Commentary on Lake et al., Behavioral and Brain Sciences, 2017
- Learning model-based strategies in simple environments with hierarchical q-networks
- Deep Episodic Value Iteration for Model-based Meta-Reinforcement Learning
- Modular Continual Learning in a Unified Visual Environment
- General AI Challenge - Round One: Gradual Learning
- A Useful Motif for Flexible Task Learning in an Embodied Two-Dimensional Visual Environment