Comparison of Neural Network Models Applied to Human Recognition

Abstract:

In this paper a comparison among conventional Artificial Neural Networks (ANN), Ensemble Neural Networks (ENN) and Modular Granular Neural Networks (MGNNs) is performed. This comparison is performed use 10-fold cross-validation using from 1 to 12 images for the training phase. Some parameters of neural networks are randomly established such as: the number of sub modules (for ensemble and modular granular neural networks), the number of neurons of two hidden layers for each sub module and learning algorithm. A benchmark database is used to observe the neural networks performances.

Año de publicación:

2021

Keywords:

Fuente:

google