Adaptive support vector machine with homogeneous decision function

In this paper we propose a new algorithm called adaptive support vector machine with homogeneous decision function. In our algorithm, the distribution of samples has been taken into consideration, so that the margin of bounding hyperplanes is as large as possible. Moreover, we introduce a pair of parameters v₊ and v_- to control bounds of the fractions of support vectors and margin errors. We also show that our algorithm can deal with imbalanced data effectively. Experiments on several artificial and UCI datasets indicate the proposed algorithm has good classification accuracy.