A Novel Algorithm for Predicting β-barrel Outer Membrane Proteins Using ACO-based Hyper-parameter Selection for LS-SVMs

摘要

An ACO-based hyper-parameter selection for least squares support vector machines (LS-SVMs) was trained to predict the topology of transmembrane β strands proteins. It should be stressed that it is very important to do a careful model selection of the tuning parameters for LS-SVM. In this paper, a novel hyper-parameter selection method for LS-SVMs is presented based on the ant clony optimization (ACO). Optimal LS-SVMs parameters for RBF kernel are selected to predict the topology of the transmembrane β strands proteins. The feasibility of this method is examined on one test database set. For the testing database, the present LS-SVMs method with RBF kernel predicts higher accuracy than SVM and HMM method. The simulation result shows that this prediction model for transmembrane β strands proteins is accurate.