FAST SUPPORT VECTOR MACHINE (FSVM) CLASSIFIER FOR STREAMING DATA

摘要

The present subject matter relates to a method and a system (200) for developing a Fast Support Vector Machine (FSVM) classifier using a Support Vector Machine (SVM) classifier (212). In an embodiment, an optimization technique is applied to the SVM classifier (212) to result in the FSVM classifier. The optimization technique includes obtaining an Acceptable Recognition Accuracy (ARA) and a set of support vectors by training and testing the SVM classifier (212) by initializing zero value to the threshold Lagrange Multiplier. Next, the technique includes the updating the value of the threshold Lagrange Multiplier and repeating the training and testing till a determined recognition accuracy (RA) is greater than or equal to the Acceptable Recognition Accuracy (ARA). Further, in an embodiment, the present subject matter relates to a real time speech recognition system implemented with the FSVM classifier