Comparison of Fully Connected Net with Particle Swarm Optimization Neural Network and PSO in the Diagnosis of Heart

Abstract

This paper proposes an Enhanced Hybrid Particle Swarm Optimization (PSO) with Artificial Neural Network (ANN), which is applied in the diagnosis of heart disease of the common features in University of California, Irvine (UCI) dataset. This UCI data includes 303 test results and consist of 13 features with two classes. One class is with health people and the other class of people are with heart disease. PSO-ANN combined Particle Swarm Optimization (PSO) and Artificial Neural Network (ANN), using ANN`s escaping mechanism to enhance the deficiency of PSO slow convergence and easy to fall into the local optimal solution. The overall search ability is increased and the tracking time is reduced. This paper uses fully connected net with PSO-ANN with Python environment compares with PSO in R, the result demonstrates that the proposed model is better than PSO around 12%.