conftrace_

2009 NeurIPS NeurIPS 2009

Skill Discovery in Continuous Reinforcement Learning Domains using Skill Chaining

Abstract

We introduce skill chaining, a skill discovery method for reinforcement learning agents in continuous domains, that builds chains of skills leading to an end-of-task reward. We demonstrate experimentally that it creates skills that result in performance benefits in a challenging continuous domain.

🌉 Interdisciplinary Bridge - Artificial Intelligence and Reinforcement Learning

📈 Trend Setter - Robotics

🐣 Hot Topic Early Bird - continuous control

Authors

George Konidaris, Andrew G. Barto

Topics

Artificial Intelligence > Core AI > Agent Systems Reinforcement Learning > Methods > Deep RL Reinforcement Learning > Methods > Policy Learning Reinforcement Learning > Applications > Robotics Artificial Intelligence > Core AI > Robotics Artificial Intelligence > Core AI > Reinforcement Learning

Keywords

reinforcement learning hierarchical learning continuous control skill discovery skill chaining continuous domain

More from NeurIPS 2009

All NeurIPS 2009 papers →

Grouped Orthogonal Matching Pursuit for Variable Selection and Prediction

Rank-Approximate Nearest Neighbor Search: Retaining Meaning and Speed in High Dimensions

Online Learning of Assignments

Strategy Grafting in Extensive Games

Training Factor Graphs with Reinforcement Learning for Efficient MAP Inference

Learning a Small Mixture of Trees