Rule-based Evolutionary Online Learning Systems

by Butz, Martin V.

• ISBN13: 9783540253792
• ISBN10: 3540253793
• Format: Hardcover
• Copyright: 2006-01-15
• Publisher: Springer-Verlag New York Inc

Summary

"This book provides a comprehensive introduction to learning classifier systems (LCS) - or, more generally, rule-based evolutionary online learning systems. While an LCS learns interactively, much like a neural network, its rule based structure increases adaptability and flexibility. The book offers a principled, modular analysis approach to understand, analyze, and design LCSs. Necessary background knowledge on problem types, genetic algorithms, and reinforcement learning is provided. The analysis is carried out on the XCS classifier system, currently the most prominent system in LCS research. Several improvements are also introduced to XCS. An application suite is provided including classification, reinforcement learning, function approximation, and datamining problems. In accordance to the originally envisioned cognitive systems, proposed by the creator of LCSs, John Holland, the book closes by outlining the potentials of LCSs for the design of competent artificial cognitive systems."--BOOK JACKET.

Table of Contents

Introduction

Prerequisites

Simple Learning Classifier Systems

The XCS Classifier System

How XCS Works: Ensuring Effective Evolutionary Pressures

When XCS Works: Towards Computational Complexity

Effective XCS Search: Building Block Processing

XCS in Binary Classification Problems

XCS in Multi-Valued Problems

XCS in Reinforcement Learning Problems

Facetwise LCS Design

Towards Cognitive Learning Classifier Systems

Table of Contents provided by Publisher. All Rights Reserved.