Multi-Nyström Method Based on Multiple Kernel Learning for Large Scale Imbalanced Classification

Figure 1

Architecture of the proposed multi-Nyström method. subsets from the majority class are sampled to construct balanced landmark points and then the Nyström method is used to obtain the approximate base kernel matrices and the multiple kernel learning (MKL) algorithm is applied to optimize the mixture weights and train classifier. Finally, the trained kernel classifier based on multi-Nystrom is obtained.