Paper Info
Title
Reinforcement Learning as Classification: Leveraging Modern Classifiers
Abstract
The basic tools of machine learning appear in the inner loop of most reinforcement learning al- gorithms, typically in the form of Monte Carlo methods or function approximation techniques. To a large extent, however, current reinforcement learning algorithms draw upon machine learn- ing techniques that are at least ten years old and, with a few exceptions, very little has been done to exploit recent advances in classification learn- ing for the purposes of reinforcement learning. We use a variant of approximate policy iteration based on rollouts that allows us to use a pure clas- sification learner, such as a support vector ma- chine (SVM), in the inner loop of the algorithm. We argue that the use of SVMs, particularly in combination with the kernel trick, can make it easier to apply reinforcement learning as an "out- of-the-box" technique, without extensive feature engineering. Our approach opens the door to modern classification methods, but does not pre- clude the use of classical methods. We present experimental results in the pendulum balancing and bicycle riding domains using both SVMs and neural networks for classifiers.
Year
Venue
Keywords
2003
ICML
machine learning,support vector,neural network,reinforcement learning,function approximation,monte carlo method
Field
DocType
Citations 
Online machine learning,Instance-based learning,Semi-supervised learning,Active learning (machine learning),Pattern recognition,Computer science,Unsupervised learning,Artificial intelligence,Computational learning theory,Machine learning,Reinforcement learning,Learning classifier system
Conference
70
PageRank 
References 
Authors
4.98
14
2
Name
Order
Citations
PageRank
Michail G. Lagoudakis1116479.51
Ronald Parr22428186.85