Support vector machines with evolutionary interval neural networks for granular feature transformation in making effective biomedical data classification

In this paper, we use new evolutionary interval neural networks to do granular feature transformation based on granular computing, neural computing and evolutionary computation to alleviate kernel's burden in support vector machines (SVMs) and help SVMs learn knowledge effectively. Simulation results for three different medical data sets show that SVMs using the evolutionary interval neural networks are more effective than the traditional SVMs in terms of testing accuracy.