Decision Support in Cancer Base on Fuzzy Adaptive PSO for Feedforward Neural Network Training

Abstract

In the last decade, the use of artificial neural networks (ANN) has become widely accepted in medical applications for accuracy for predictive inference, with potential to support and flexible non-linear modelling of large data sets. Feedforward neural network (FNN) is a kind of artificial neural networks, which has a better structure and been widely used. But there are still many drawbacks if we simply use feedforward neural network, such as slow training rate, easy to trap into local minimum point, and bad ability on global search. In this paper, feedforward neural network trained by fuzzy adaptive particle swarm optimization (FPSO) algorithm is proposed for breast cancer diagnosis. The results which are compared with FNN trained by PSO algorithm show much more accurate and stable, converges quickly towards the optimal position and can avoid overfitting in some extent. In the computer-aided decision systems, the accurate and stable computer algorithms are very important to help a physician in diagnosing a patient.