Kernel K-local Hyperplanes for Predicting Protein-Protein Interactions

Protein-protein interactions prediction is an important problem in biology. In this paper, kernel method is coupled with HKNN to develop a new method, kernel k-local hyperplanes (KHKNN), to predict Protein-protein interactions. The main idea behind KHKNN is to first map the input into a higher-dimensional feature space with some non-linear transformation, which is implicitly induced by a predefined kernel and then to train a HKNN classifier there rather than in the original input space. Moreover the introduction of kernel function makes KHKNN freely to be used for the specific application problem. Experimental results have demonstrated that KHKNN is a useful method for the prediction of protein-protein interactions and can be used to other classifying tasks.