Genetic Programming with Architecture - Altering Operations
نویسندگان
چکیده
The goal of automatic programming is to create, in an automated way, a computer program that enables a computer to solve a problem. Ideally, an automatic programming system should require that the user pre-specify as little as possible about the problem environment. In particular, it is desirable that the user not be required to prespecify the architecture of the ultimate solution to his problem. The question of how to automatically create the architecture of the overall program in an evolutionary approach to automatic programming, such as genetic programming, has a parallel in the biological world: how new structures and behaviors are created in living things. This corresponds to the question of how new DNA that encodes for a new protein is created in more complex organisms. This chapter describes how the biological theory of gene duplication described in Susumu Ohno's provocative book, Evolution by Means of Gene Duplication, was brought to bear on the problem of architecture discovery in genetic programming. The resulting biologicallymotivated approach uses six new architecture-altering operations to enable genetic programming to automatically discover the architecture of the solution at the same time as genetic programming is evolving a solution to the problem. Genetic programming with the architecture-altering operations is used to evolve a computer program to classify a given protein segment as being a transmembrane domain or non-transmembrane area of the protein (without biochemical knowledge, such as the hydrophobicity values used in human-written algorithms for this task). The best genetically-evolved program achieved an out-of-sample error rate that was better than that reported for other previously reported human-written algorithms. This is an instance of an automated machine learning algorithm matching human performance on a non-trivial problem.
منابع مشابه
Evolving the Architecture of a Multi-part Program in Genetic Programming Using Architecture-Altering Operations
This paper describes six new architecture-altering operations that provide a way to dynamically determine the architecture of a multipart program during a run of genetic programming. The new operations are patterned after the naturally occurring operations of gene duplication and gene deletion and are motivated by Ohno's provocative book Evolution by Means of Gene Duplication. The new operation...
متن کاملUse of Automatically Defined Functions and Architecture- Altering Operations in Automated Circuit Synthesis with Genetic Programming
This paper demonstrates the usefulness of automatically defined functions and architecture-altering operations in designing analog electrical circuits using genetic programming. A design for a lowpass filter is genetically evolved in which an automatically defined function is profitably reused in the 100% compliant circuit. The symmetric reuse of an evolved substructure directly enhances the pe...
متن کاملClassifying Protein Segments as Transmembrane Domains Using Architecture-Altering Operations in Genetic Programming
متن کامل
Evolution of Both the Architecture and the Sequence of Work-Performing Steps of a Computer Program Using Genetic Programming with Architecture-Altering Operations
The goal of automatic programming is to create, in an automated way, a computer program that enables a computer to solve a problem. Ideally, an automatic programming system should require that the user pre-specify as little as possible about the problem environment. In particular, it is desirable that the user not be required to prespecify the architecture of the ultimate solution to his proble...
متن کاملGene Duplication to Enable Genetic Programming to Concurrently Evolve Both the Architecture and Work-Performing Steps of a Computer Program
Susumu Ohno's provocative book Evolution by Gene Duplication proposed that the creation of new proteins in nature (and hence new structures and new behaviors in living things) begins with a gene duplication and that gene duplication is "the major force of evolution." This paper describes six new architecture-altering operations for genetic programming that are patterned after the naturally-occu...
متن کامل