Research on Brain Signals via Artificial Neural Network and Swarm Intelligence Algorithms

Abstract

Artificial Neural Networks (ANNs) that are the ability to learn from their environment in order to improve their performance are widely used in numerous applications. The Backpropagation (BP) Algorithm is one of the most popular and effective model of ANNs. However, since it uses gradient descent algorithm, which attempts to minimize the error of the network by moving gradient of the error curve, easily get trapped at local minima. In order to avoid this problem and to obtain a better classifier, we proposed an ANNs and Swarm Intelligence (SI) method where Artificial Bee Colony and Particle Swarm Optimization algorithms were operated for the Multilayer Perceptron Neural Network. Two Electroencephalogram (EEG) datasets were used to test to test the accuracy and success of the study performed. Compared to conventional-MLPNN, higher success values were obtained on each dataset with the proposed methods. Experimental results demonstrate that combined SI and MLPNN algorithm has been increased the success of BP algorithm by avoiding local minima. For ABC data, respectively, ABC-MLPNN and PSO-MLPNN methods, 79.00% and 75.50% respectively for Boston data and 91.67% and 88.33% respectively for Selcuk data were obtained. On the other hand, with the MLPNN algorithm, the success rate was 68.50% for Boston data and 81.67% for Selcuk data. These results show that the success of MLPNN algorithm significantly increases with the weights obtained by using SI. In addition to this, this study showed that the SI-MLPNN algorithm can be used on non-linear and highly complex EEG data.