Soft Computing Techniques based Recursive Error Correcting Output Code for Multi-Class Pattern Classification

Abstract

The pattern recognition methods like speech recognition, text classification, image recognition results in the solving of multi-class problems. This can be achieved by means of classifying multi-class problems into several two class problems using the Soft Computing techniques such as Neural Networks and Support Vector Machines. The best code matrix for a given problem cannot be designed taking into account only the features of the code matrix, viz., overall classifier accuracy, minimum hamming distance and margin of classification, but also the features of the problem, viz., attributes, samples and classes are to be considered. Conventionally, code matrix is designed based on either the features of the problem or the features of the code matrix. The proposed work, focused on designing a new code matrix based on both the features of the problem and code matrix. In order to improve the accuracy and reduce computation time, the generation of feature dependent code matrix through an evolutionary algorithm is proposed. This model aims at developing a hybrid version of Recursive Error Correcting Output Code with Biogeography Based Optimization to achieve maximum classification accuracy and minimum computational time.