TrIK-SVM : an alternative decomposition for kernel methods in Krein spaces
This work addresses a specific computational bottleneck in kernel methods for researchers in machine learning, but it appears incremental as it builds on existing KSVM approaches.
The authors tackled the problem of kernel machines with indefinite kernels by proposing an alternative kernel decomposition that avoids eigen-decomposition, resulting in a method that does not require computing the full kernel matrix and shows good behavior compared to KSVM.
The proposed work aims at proposing a alternative kernel decomposition in the context of kernel machines with indefinite kernels. The original paper of KSVM (SVM in Kreǐn spaces) uses the eigen-decomposition, our proposition avoids this decompostion. We explain how it can help in designing an algorithm that won't require to compute the full kernel matrix. Finally we illustrate the good behavior of the proposed method compared to KSVM.