Filter Size Optimization on a Convolutional Neural Network Using FGSA

Abstract

This paper presents an approach to optimize the filter size of a convolutional neural network using the fuzzy gravitational search algorithm (FGSA). The FGSA method has been applied in others works to optimize traditional neural networks achieving good results; for this reason, is used in this paper to optimize the parameters of a convolutional neural network. The optimization of the convolutional neural network is used for the recognition and classification of human faces images. The presented model can be used in any image classification, and in this paper the optimization of convolutional neural network is applied in the CROPPED YALE database.