F. Camacho, D. Manrique, and A. Rodríguez-Patón (Spain)
Genetic Algorithms, Radial Basis Function Networks, Adaptive Intelligent Systems, Breast Cancer Diagnosis.
This paper presents an evolutionary system for automatically designing radial basis function networks. The work presented here allows the construction of self adaptive intelligent systems based on radial basis function networks. It consists of a genetic algorithm that employs a new binary codification of radial basis architectures and a crossover operator especially designed to take advantage of the proposed codification. The goal of the genetic system is to output a neural architecture that better solves a problem described as a set of training patterns. This system has been applied to a real-world classification problem: breast cancer diagnosis, using a data base of patients taken from a Spanish hospital. The results show that the overall performance of the proposed system is better than other related approaches with which it has been compared.
Important Links:
Go Back
