Diagnosis of Diabetes using Modified PSO and FFNN Technique Their Application to WSD

Abstract

It is one of the most amazing and perplexing feats of human mind that we understand written and spoken communication in spite of enormous number of possibilities that exist because of multiple meanings of words that compose a sentence. Human beings can produce a correct sentence choosing words in their appropriate context with little effort. But the same problem becomes very hard and complex when it is sought to be automated. Therefore any system that proposes to implement Natural Language Processing (NLP) on a computer has to address very seriously the question of WSD. The most important application of WSD is diagnosis of diabetes. This paper is implementation of diagnosis of diabetes using PSO and FFNN. To classify the diabetic detection, we used a feed forward network (FFNN) in which the network can be trained to optimize the mean square error (MSE) by using a modified particle swarm optimization (PSO) algorithm. The effectiveness for this procedure was verified by an experimental analysis on a benchmark dataset which is publicly available on UCI learning repository. The result of our experimental analysis revealed that the improvement on the algorithm is significant with respect to FF trained by PSO. Using these attribute to Neural Network as input we have achieved the best known training classification accuracy of 76.0417%.