Predicting Drug-target Interaction using Support Vector Machine and Invasive Tumor Growth Optimization

Abstract

Prediction of drug-target interaction is a core problem in drug discovery. In these years, more machine learning methods have been used to solve this problem, but invalid due to the imbalanced data set. In this paper, we propose a new ensemble learning framework by the support vector machine(SVM) and invasive tumor growth optimization(ITGO) algorithm. ITGO is used to solve the penalty parameter optimization problem of SVM for the imbalanced data set. In order to verify the performance of our methods, four benchmark dataset are chosen to compare with the well-known methods. Experimental results show that our method has better effectiveness and robustness than other methods.