Weighted Core Vector Machine

Classification for large datasets is a classical problem in machine learning. Thegeneralized core vector machine can deal with large datasets problem. but it did not take intoaccount the distribution of the data set of sample points, the error term for all sample pointsare given the same punishment factor,the classifier effect is very susceptible.To address large datasets and imbalanced datasets two issues, we propose the weightedcore vector machine algorithm, which mainly includes the following two aspects:1、A fast learning algorithm on large datasets called a core set weighted support vectormachines is proposed.In step one,in the proposed approach, the corresponding core set can besolved by employing the core vector machine (CVM) or generalized CVM (GCVM). In steptwo, the weighted support vector machines (WSVM) can be used to implement classificationfor imbalanced datasets.2、A new classification approach is proposed in order to deal with large size andimbalanced datasets classification problem. The proposed method is based on maximumvector angular margin core vector machine to implement classification for large datasets. Forthe imbalanced datasets classification problem, each sample is assigned by different weights,which can improve the classification performance of algorithm. The proposed approach caneffectively solve the imbalanced datasets classification problem and implement fast trainingon large datasets.The two methods is used in UCI data set, and it is compared with other methodsrespectively. The experimental results show that the weighted core vector machine andweighted maximum vector angular margin core vector machine have better classificationperformance.