A higher order evolutionary Jordan Pi-Sigma Neural Network with gradient descent learning for classification
Janmenjoy Nayak, Devi Prasanna Kanungo, Bighnaraj Naik, Himansu Sekhar Behera · 2014
Solving nonlinear classification problems by the traditional neural networks with one or more hidden units is a quite tough task. Various researchers around the globe have made tremendous effort to solve such problem with the help of some Higher Order Neural Networks (HONN). The capability of expanding the input space of HONN makes them more efficient for solving complex problems. In this paper, a standard back propagation gradient descent learning trained higher order Jordan Pi-Sigma Neural Network (JPSNN) has been proposed for classification of real world data. For optimizing the performance of the network, genetic algorithm has been coupled with JPSNN and the resulting performance has been compared with another HONN called Pi-Sigma network. The proposed method has been tested with various real world benchmark datasets considered from UCI machine learning dataset repository and the simulated results are being tested with the statistical tool ANOVA to indicate the statistical significance of our result.