Semi-Supervised Learning Machine Based On Multi-View Twin Support Vector Machine

Semi-supervised learning problem is one of the most common learning problems in machine learning.Not only does it come with labeled data,but it also comes with a lot of unlabeled data.It exists in reality such as spam filtering,medical image analysis and syntactic analysis.The multi-view semi-supervised learning algorithm combines multi-view with semi-supervised learning,which is better than single-view classification.In this paper,we proposed a method combined multi-view semi-supervised learning with the twin support vector machine.The specific content as follows:For classification problem,when the data set has different features,the data is divided into multi-views according to different features.The model is constructed by using the twin support vector machine to find the non-parallel hyperplanes by solving two small-scale quadratic programming problems,so that hyperplanes are as close to corresponding sample points as possible,but away from another sample points.Meanwhile the unlabeled data points can be labeled.For nonlinear partitioning problems,this paper construct introducing ptimal model by kernel function.In this paper,the multi-view twin support vector machine semi-supervised learning method is tested by artificial data set and UCI data set respectively.The experimental results show that,compared with the twin support vector machine,the running time of this method is shortened,the computational complexity is reduced,the classification accuracy is good,and the prediction performance is better.