Abstract | ||
---|---|---|
This paper discusses parameter-based exploration methods for reinforcement learning. Parameter-based methods perturb parameters of a general function approximator directly, rather than adding noise to the resulting actions. Parameter-based exploration unifies reinforcement learning and black-box optimization, and has several advantages over action perturbation. We review two recent parameter-exploring algorithms: Natural Evolution Strategies and Policy Gradients with Parameter-Based Exploration. Both outperform state-of-the-art algorithms in several complex high-dimensional tasks commonly found in robot control. Furthermore, we describe how a novel exploration method, State-Dependent Exploration, can modify existing algorithms to mimic exploration in parameter space. |
Year | DOI | Venue |
---|---|---|
2010 | 10.2478/s13230-010-0002-4 | Paladyn |
Keywords | Field | DocType |
reinforcement learningoptimizationexplorationpolicy gradients,parameter space,evolution strategy,generating function,reinforcement learning,robot control | Robot control,Computer science,Simulation,Artificial intelligence,Parameter space,Machine learning,Reinforcement learning | Journal |
Volume | Issue | ISSN |
1 | 1 | 2080-9778 |
Citations | PageRank | References |
24 | 1.17 | 19 |
Authors | ||
6 |
Name | Order | Citations | PageRank |
---|---|---|---|
Thomas Rckstieß | 1 | 24 | 1.17 |
Frank Sehnke | 2 | 527 | 39.18 |
Tom Schaul | 3 | 916 | 79.40 |
Daan Wierstra | 4 | 5412 | 255.92 |
Yi Sun | 5 | 74 | 10.99 |
Jürgen Schmidhuber | 6 | 17836 | 1238.63 |