Gene expression programming for symbolic regression

• Main Author: Jandhyala Manognya
• Other Authors: Wang Lipo
• Format: Final Year Project
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: http://hdl.handle.net/10356/17859
• Institution: Nanyang Technological University

Gene Expression Programming is an evolutionary algorithm that mimics biological evolution to solve a user defined problem. Just like chromosomes are used to represent human beings, each chromosome in the GEP seeks to represent the solution to the problem. GEP uses linear character chromosomes made up of genes, each of which contains a head and a tail. Population of chromosomes are created to best fit a selection environment. With repeated modification or evolution by means of mutation, inversion, transposition and reproduction, the perfect solution to problems can be achieved. Many problems can be solved to illustrate the power and versatility of gene expression programming. Some of the more famous ones include symbolic regression, decision tree induction, designing of neutral networks, combinational optimization, etc. That said, this report would be focusing on solving the problem of symbolic regression and induction state transducers. The term symbolic regression is the process by which a set of data is made to fit to a mathematical formula. This process is very frequently used in experiments as the experimental results always seem to point to a pattern or relationship between the variables in the experiment. Transducers are state machines which depict a behaviour or pattern of a certain system. Although there have been solution to state machines, GEP was not used before. The author thus, proposes a method to using GEP to solve Induction state transducer problems.