Research on Multi-View Semi-Supervised Learning Algorithm Based on Co-Learning

Recent years multi-view semi-supervised learning has become research focus. In most cases multiple views are often supposed to be given previous to learning. However it is not the case in the real-world application, which makes multi-view semi-supervised learning algorithms impractical and infeasible. A view partitioning method called ViewPartition was proposed. It´s used to partition input features into two parts. Based on ViewPartition, a new multi-view semi-supervised learning algorithm called Co-VP was presented. Co-VP can construct classifiers from labeled and unlabeled data. Studies comparing classification algorithms have found Co-VP to be comparable in performance with classification trees and with neural network classifiers. They have also exhibited high accuracy when applied to real-world databases, especially for those with more redundant features