Hierarchical Imitation and Reinforcement Learning

Abstract

We study how to effectively leverage expert feedback to learn sequential decision-making policies. We focus on problems with sparse rewards and long time horizons, which typically pose significant challenges in reinforcement learning. We propose an algorithmic framework, called hierarchical guidance, that leverages the hierarchical structure of the underlying problem to integrate different modes of expert interaction. Our framework can incorporate different combinations of imitation learning (IL) and reinforcement learning (RL) at different levels, leading to dramatic reductions in both expert effort and cost of exploration. Using long-horizon benchmarks, including Montezuma's Revenge, we demonstrate that our approach can learn significantly faster than hierarchical RL, and be significantly more label-efficient than standard IL. We also theoretically analyze labeling cost for certain instantiations of our framework.

arXiv commentProceedings of the 35th International Conference on Machine Learning (ICML 2018)

References in corpus (10)

• Prioritized Experience Replay
• Dueling Network Architectures for Deep Reinforcement Learning
• Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivation
• Deep reinforcement learning from human preferences
• Deep Q-learning from Demonstrations
• Modular Multitask Reinforcement Learning with Policy Sketches
• Why We Need New Evaluation Metrics for NLG
• Reinforcement and Imitation Learning via Interactive No-Regret Learning
• Learning to Search Better Than Your Teacher
• Deep Abstract Q-Networks

Cited by in corpus (9)

• Beyond spiking networks: the computational advantages of dendritic amplification and input segregation
• Watch, Try, Learn: Meta-Learning from Demonstrations and Reward
• Randomized Value Functions via Multiplicative Normalizing Flows
• Hindsight Trust Region Policy Optimization
• Online Baum-Welch algorithm for Hierarchical Imitation Learning
• Efficient Robotic Object Search via HIEM: Hierarchical Policy Learning with Intrinsic-Extrinsic Modeling
• Reparameterized Variational Divergence Minimization for Stable Imitation
• Discovering hierarchies using Imitation Learning from hierarchy aware policies
• Active Hierarchical Imitation and Reinforcement Learning