Global+Optimization+Algorithms+Theory+and+Application_Part6

Global+Optimization+Algorithms+Theory+and+Application_Part6...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 2.1 Introduction 101 4. Learning Classifier Systems (LCS), discussed in Chapter 7 on page 233 , are online learning approaches that assign output values to given input values. They internally use a genetic algorithm to find new rules for this mapping. 5. Evolutionary programming (EP, see Chapter 6 on page 231 ) is an evolutionary approach that treats the instances of the genome as different species rather than as individuals. Over the decades, it has more or less merged into Genetic Programming and the other evolutionary algorithms. Evolutionary Algorithms Evolutionary Programming Evolution Strategy Differential Evolution Genetic Algorithms Learning Classifier Systems Genetic Programming GGGP LGP SGP Figure 2.2: The family of evolutionary algorithms. The early research [518] in genetic algorithms (see Section 3.1 on page 141 ), Genetic Programming (see Section 4.1.1 on page 157 ), and evolutionary programming (see Section 6.1 on page 231 ) date back to the 1950s and 60s. Besides the pioneering work listed in these sections, at least other important early contribution should not go unmentioned here: The Evolutionary Operation (EVOP) approach introduced by Box [260], Box and Draper [261] in the late 1950s. The idea of EVOP was to apply a continuous and systematic scheme of small changes in the control variables of a process. The effects of these modifications are evaluated and the process is slowly shifted into the direction of improvement. This idea was never realized as a computer algorithm, but Spendley et al. [1941] used it as basis for their simplex method which then served as progenitor of the downhill simplex algorithm 11 of Nelder and Mead [1517]. [518, 1276] Satterthwaite’s REVOP [1815, 1816], a randomized Evolutionary Operation approach, however, was rejected at this time [518]. We now have classified different evolutionary algorithms according to their semantics, in other words, corresponding to their special search and problem spaces. All five major approaches can be realized with the basic scheme defined in Algorithm 2.1 . To this simple structure, there exist many general improvements and extensions. Since these normally do not concern the search or problem spaces, they also can be applied to all members of the EA family alike. In the further text of this chapter, we will discuss the major components of many of today’s most efficient evolutionary algorithms [357]. The distinctive features of these EAs are: 1. The population size or the number of populations used. 11 We discuss Nelder and Mead [1517]’s downhill simplex optimization method in Chapter 16 on page 283 . 102 2 Evolutionary Algorithms 2. The method of selecting the individuals for reproduction....
View Full Document

This document was uploaded on 08/10/2011.

Page1 / 20

Global+Optimization+Algorithms+Theory+and+Application_Part6...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online