Laplacian Lp norm least squares twin support vector machine

Semi-supervised learning has become a hot learning framework, where large amounts of unlabeled data and small amounts of labeled data are available during the training process. The recently proposed Laplacian least squares twin support vector machine (Lap-LSTSVM) is an excellent tool to solve the semisupervised classification problem. Motivated by the success of Lap-LSTSVM, in this paper, we propose a novel Laplacian Lp norm least squares twin support vector machine (Lap-LpLSTSVM). There are several advantages of our proposed method: (1) The performance of our proposed Lap-LpLSTSVM can be improved by the adjustability of the value of p. (2) The introduced Lp norm graph regularization term can efficiently exploit the geometric information embedded in the data. (3) An efficient iterative strategy is employed to solve the optimization problem. Besides, to demonstrate that our proposed method can make use of unlabeled data effectively, least squares twin support vector machine (LSTSVM) which only uses the same labeled data is used to compare with our proposed method. The experimental results on both synthetic and real-world datasets show that our proposed method outperforms other state-of-theart methods and can also deal with noisy datasets.

Automated summary

In this paper, a novel semi-supervised learning method called Lap-LpLSTSVM is proposed, which utilizes Lp norm least squares twin support vector machine to handle classification problems. The method has the advantages of adjustable performance, efficient utilization of geometric information, and effective optimization, enabling the use of unlabeled data and handling of noisy datasets.