Improving Performance of k-NN Classifier using Prototype Generation: Survey

Abstract

Databases contain huge amount of hidden information that can be used for intelligent decision making. Classification is one of the forms of data analysis that can be used to extract models that represents classes of data. Although k-NN is well known pattern classification algorithm used in different applications, it has some loopholes for large dataset. First limitation is, it requires whole training set to be stored in memory. Also for classifying a test pattern it has to be compared with all other training instances. So in order to improve the performance of k-NN classifier prototype generation are used. Prototype reduction techniques can be divided into two different approaches, known as prototype selection and prototype generation or abstraction. By building new artificial prototypes, prototype generation increase accuracy of NN (Nearest Neighbor) classification. EMOPG provides better results in terms of accuracy and reduction than other Prototype Generation methods.