Biological data analysis using hybrid functional link artificial neural network

Abstract

With rapid growth in field of medical sciences; biologists have stepped up and had discovered many biological characteristics for different diseases. There are many issues involved in knowledge discovery from these huge biomedical data for useful medical application. Traditional algorithms for neural network such as multi-layer perceptron model have proven to be inefficient for the classification of different biological data, as in many cases to get a better classifier we have to increase number of layers in simple artificial neural network which increases the complexity of the network. In contrast to multiple layer perceptron networks, Functional Link Artificial Neural Network (FLANN) can be implemented for the task of data classification with reduced complexity. In this proposed paper work an experimental study has been presented where a simple FLANN based classification model is compared with Hybrid FLANN models. In the first hybrid model an optimization technique i.e. Particle Swarm Optimization (PSO) is applied along with FLANN for weight updation. Then in the second hybrid FLANN model a feature selection technique i.e. Signal to Noise Ratio (SNR) is applied. Finally the classification accuracies are compared for the estimation of good performances.