Density Weighted Core Support Vector Machine

Core Vector Machine (CVM) can be used to deal with large datasets classification problem, but CVM do not consider the densitydistribution of the data. In order to obtain the optimal descriptionof the data, we propose a density weighted core support vectormachine (DWCVM). In the proposed DWCVM, the relativedensity of each data point is based on the density distribution ofthe target data using the k-nearest neighbor (k-NN) approach.Experimental results on several benchmark data sets show thatthe performance of DWCVM is much better than CVM.

Density Weighted Core Support Vector Machine Keywords:

minimum enclosing ball , core set , support vectordomaindescription , density , core vector machine