Advancements in Multi-Layer Perceptron Training to Improve Classification Accuracy

Abstract

Neural Networks are the popular classification tools used in Medical diagnosis for early disease detection. The performance of Neural Networks is highly depended on the training process. In the training process, the individual weights between each of the neuron are adjusted for better classification results. Many Gradient-based and Meta-heuristic training algorithms are proposed and used by the researchers to improve the training performance of Neural Network. However, there are some limitations in both Gradient-based and Meta-heuristic algorithms when there are used individually. To overcome these limitations and to improve the Multi-Layer Perceptron Network performance Hybrid algorithms are useful. In this study, a review on advancements in Multi-Layer Perceptron Network training process for the improvement of classification performance is presented.