|
Buy the
Book
Home
News
Author
Q&A
Tutorials
Downloads
GEP Biblio
Contacts
Visit Gepsoft
|
|
|
|
|
Gene Expression Programming: Mathematical Modeling by an Artificial Intelligence
|
|
| Preface |
| |
I developed the basic ideas of gene expression programming (GEP) in September and October of 1999 almost unaware of their uniqueness. I was reading Mitchell’s book
An Introduction to Genetic Algorithms (Mitchell
1996) and meticulously solving all the computer exercises provided at the end of each chapter. Therefore, I implemented my first genetic algorithm and I also implemented what I thought was a genetic programming system (GP). Like a GP system, this new system could also evolve computer programs of different sizes and shapes but, surprisingly, it surpassed the old GP system by a factor of 100-60,000. So, what happened here? What was responsible for this astounding difference in performance? For an evolutionary biologist, the answer is quite straightforward: this new system – gene expression programming – simply crossed the phenotype threshold. This means that the complex computer programs (the phenotype) evolved by GEP are totally encoded in simple strings of fixed length (the chromosomes or genotype). The separation of the genotype from the phenotype is comparable to opening a Pandora box full of good things or possibilities. Of these good things, perhaps the most important is that there are virtually no restrictions concerning the number or type of genetic operators used. Another important thing is that the creation of higher levels of complexity becomes practically a trivial task. Indeed, it was trivial to create a multigenic system from a unigenic one and a multicellular from a unicellular. And each new system creates its own box of new possibilities, which enlarge considerably the scope of this new technique.
In this first book on gene expression programming I describe thoroughly the basic gene expression algorithm and numerous modifications to this new algorithm, providing all the implementation details so that anyone with elementary programming skills (or willing to learn them) will be able to implement it themselves. The
first chapter briefly introduces the main players of biological gene expression in order to show how they relate to the main players of artificial evolutionary systems in general and GEP in particular. The
second chapter introduces the players of gene expression programming, showing their structural and functional organization in detail. The language especially created to express the genetic information of GEP chromosomes is also described in this chapter.
Chapter 3 gives a detailed description of the basic gene expression algorithm and the basic genetic operators. In addition, a very simple problem is exhaustively dissected, showing all the individual programs created during the discovery process in order to demystify the workings of adaptation and evolution.
Chapter 4 describes some of the applications of the basic gene expression algorithm, including a large body of unpublished materials, namely, parameter optimization, evolution of Kolmogorov-Gabor polynomials, time series prediction, classifier systems, evolution of linking functions, multicellularity, automatically defined functions, user defined functions and so forth. The materials of
Chapter 5 are also new and show how to simulate complete neural networks with gene expression programming. Two benchmark problems are solved with these GEP-nets, providing an effective measure of their performance.
Chapter 6 shows how to do combinatorial optimization with gene expression programming. Multigene families and several combinatorial-specific operators are introduced and their performance evaluated on two scheduling problems. The
last chapter discusses some important and controversial evolutionary topics that might be refreshing to both evolutionary computists and evolutionary biologists.
|
|
|
Home
|
Contents
| Next
|
|