A weight learning technique for cursive handwritten text categorization with fuzzy confusion matirx

Abstract

A fuzzy confusion matrix based cursive handwritten text categorization has been implemented. Printed text is obtained from handwritten text through Modified Optimal Clustering Algorithm (MOCA). Optimal Clustering Algorithm (OCA) groups texts into different subject categories. Learning is conducted to extract the attributes along with corresponding weights for each subjects. Fuzzy confusion matrix has been used to measure several performance metrics with Holdout method. These are satisfactory. Over and above the text learning and recognition time is very less making the system efficient also.