A Method to Decrease Data Calssifcation with Weighting Technique in SVM+

نوع مقاله : مقاله پژوهشی

نویسندگان
Technical college of Payame Noor, Tehran, Iran
چکیده
 SVM is a learning algorithm related to data analysis and detection of patterns that used for classification. But the important thing is that duplicate data and real-time processing has not been correctly calculated. That's why, we have presented a method for decreasing of dataclassification using weighting techniques in SVM+. In Comparison with SVM+ this proposed approach has optimal response time. We could divide the range size by observing the parameter data volumes and density with this classification in comparison with the examined case reduces running
time of SVM+ algorithm.Also with the objective function of the proposed method of integration of the parameters and partition data we were
able to reduce, data duplication in comparison with SVM+ and eventually we define threshold detector to qualification to reduce processing time and increase speed of data processing. Finally, the proposed algorithm with
the techniques of the SVM + weighting function is optimized in terms of efficiency.
 

کلیدواژه‌ها