Скачать или смотреть Iterative Structure-Based Grammatical Evolution

This is a video presentation of my Research Project for my honours degree in Computer Science at the University of Pretoria (COS700).

Abstract:
Genetic Programming, for all its success, often produces sub-optimal solutions. This is because individuals converge to local search areas that are difficult to escape and may not contain global optima. Structure-based algorithms have proven efficient at avoiding premature convergence. In particular, the Iterative Structure-Based Algorithm has shown success in escaping local optima and producing better results than canonical Genetic Programming. While research in structure-based genetic programming is not extensive, it is even less so in Grammatical Evolution. This paper investigates the effect of structure in Grammatical Evolution by proposing the Iterative Structure-Based Grammatical Evolution algorithm. The Iterative Structure-Based Grammatical Evolution is shown to outperform standard Grammatical Evolution on three benchmarks, successfully escaping local optima.