An Introduction to Genetic Algorithms - The MIT Press
An Introduction to Genetic Algorithms - The MIT Press is an oldie, but goodie. The official text for the University of Edinburgh Informatics course on Genetic Algorithms and Genetic Programming (being taught by Gillian Hayes in Autumn, 2006), it's already got me thinking about how the pairing of candidate chromosomes for the next generation might be achieved as it is in biological systems - i.e. not randomly, but on the basis of some gross characteristics.
Update (2006-10-17): Chapter 4 discusses what problems GAs are good at (and how they find the solution space, using the building blocks hypothesis). This got me thinking of whether Evolution is a good GA problem to solve or not. It might be at a later stage, when the fitness landscape is smooth and crossovers between two good solutions are likely to lead to a better solution. But early on in the evolutionary landscape, it is possible that GAs were NOT the best way of finding a solution (given that there was less complexity/organization in the organisms, hence crossover solutions were less likely to make sense).
Update (2006-10-17): Chapter 4 discusses what problems GAs are good at (and how they find the solution space, using the building blocks hypothesis). This got me thinking of whether Evolution is a good GA problem to solve or not. It might be at a later stage, when the fitness landscape is smooth and crossovers between two good solutions are likely to lead to a better solution. But early on in the evolutionary landscape, it is possible that GAs were NOT the best way of finding a solution (given that there was less complexity/organization in the organisms, hence crossover solutions were less likely to make sense).
posted by Sujai at 10:46
0 Comments:
Post a Comment
<< Home