Application of feature selection to data fusion based on improved perceptron and GAPSO

In this article, Relief algorithm based on feature selection has been used for data fusion since it has a better time complexity compared to rough set theory. In the other section of the article, Genetic Algorithm Particle Swarm Optimization (GAPSO) and improved Perceptron methods are used to train the network. Experiments are performed in different iterations; meanwhile the improved Perceptron and GAPSO algorithms are compared together, based on Quality of Train (QoT) and Efficiency (Ef). Improved Perceptron algorithm develops a tradeoff between test data efficiency and QoT; however, GAPSO algorithm works comparably better than Perceptron in terms of data efficiency.