Selected Methods and Algorithms for Data Classification

Abstract:

In the first part of a paper were detailed study of the discriminant analysis, based on the concept of distance between elements, and on the criterion of proximity constructed on calculating a cosine of the angle between two vectors was conducted. Also, all constructions relying on the example of texts classification are provided. In the second part of the article the method of support vector machines (SVM - Support Vector Machines), included in the set of algorithms called as "supervised learning" was examined. They are also effectively used in problems of classification as a method of discriminant analysis too. Thus, the article analyzes in detail the problem of data classification. In particular, it is proposed to modify the particle swarm optimization algorithm and use it in conjunction with the SVM algorithm.