ARCHITECTURE OPTIMIZATION MODEL FOR THE MULTILAYER PERCEPTRON AND CLUSTERING
Mohamed Ettaouil, Mohamed Lazaar, Youssef Ghanou, Ben Abdellah · 2013
This paper presents an approach called Architecture Optimization Model for the multilayer Perceptron. This approach permits to optimize the architectures for the multilayer Perceptron. The results obtained by the neural networks are dependent on their parameters. The architecture has a great impact on the convergence of the neural networks. More precisely, the choice of neurons in each hidden layer, the number of hidden layers and the initial weights has a great impact on the convergence of learning methods. In this respect, we model this choice problem of neural architecture in terms of a mixed-integer problem with linear constraints. We propose the genetic algorithm to solve the obtained model. The experimental work for classification problems illustrates the advantages of our approach.