Grant: Discovery Grants Program - Individual

The goal of the proposed research program is to study the relation between a particular method in the area of Evolutionary Computation, namely Genetic Programming (GP) , and the field of Self-Organization. The need for such a study comes from the realization that GP in its traditional form has reached a plateau of development. The technique is presently being exploited in many fields of Science and Engineering, yet its limitations become clearer every day. One key obstacle for GP is the problem of scalability, i.e. the difficulty to evolve large programs using present-day representations of programs. Another difficulty is the interaction of various non-intended, emergent phenomena of evolutionary runs with the primary goal of the artificial evolution. We aim at studying the relationship between GP and self-organization by introducing various innovative representations of programs (together with corresponding operators producing new variants of programs). Some will be non-sequential (like in our recent work on random sequences), some will be self-evolving representations, some will be otherwise complex. A representation with considerable promise is one consisting of a genome with regulatory elements (like regulatory networks in nature). Other representations will have elements of development added. The research proposed here will contribute to a vibrant research field of Computer Science, including Evolutionary Computation. In applications, the areas of Bioinformatics and Computational Finance hold many problems that are suitable for exploitation with Genetic Programming, with substantial benefits to Canada.