classifcation using multiplicative update rules in Non negative matrix factorization |
Author(s): |
| Dipali Gamit , Gujarat technological University |
Keywords: |
| Nonnegative matrix factorization (NMF), multiclass SVM classifier, static reliability measure(SRM), dynamic reliability measure( DRM), one against all approach. |
Abstract |
|
Basically nonnegative matrix factorization (NMF) is performing on the original data for data transformation (dimensionality reduction) which is followed by the SVM classification. Using multiplicative update rules, there is a classification using SVM on two classes data which employ two sets of class one belongs to maximum margin classification constraints imposed into NMF and second is belongs to maximum margin classification constraints imposed into discriminant NMF. This paper aims to multiclass classification using two approaches multiclass SVM classifier and combining two class classifier. The previous framework two class which have also a very high accuracy of classification on various database. But this paper leads keeping the classification accuracy high with doing multiclass classification which employ maximum margin between the multiple classes of data. |
Other Details |
|
Paper ID: IJSRDV2I1217 Published in: Volume : 2, Issue : 1 Publication Date: 01/04/2014 Page(s): 429-433 |
Article Preview |
|
|
|
|
