Optimization Method For LSSVM

Resolve the the Folksonomy a garbage tag issue, the garbage tag detection model. Vector space model, the characterization of user characteristics, and then support vector machine Folksonomy user two classification. Detect hidden in a normal user groups refuse to put in to reduce the waste tag number. For the bottleneck of traditional SVM algorithms in high-dimensional large-scale data sets on the speed and accuracy, and generalization is not strong features of collective emerged in a limited number of small sample data, we propose the spam label detection model based on LSSVM algorithm, experimental results garbage tag detection model based on LSSVM algorithm has higher classification accuracy and less prediction time, superior to the traditional SVM detection methods.The standard SVM algorithm, only a small part of the training data in support of the vector, thus the sparsity of good LSSVM is all the training data are support vectors, sparse poor. In addition, the ordinary LSSVM in, regardless of the location of the training points, whether there is noise pollution, the role of all the training points in the training is the same. This paper, the idea of Fuzzy SVM introduced into LSSVM for each training point defines a support to the measure, and this support is determined by the absolute value of the Lagrange factor to measure. Gradually removed some of the less important training points, using the shear method to solve the sparsity problem. Paper to experiment with the corresponding data, results show that the fuzzy LSSVM improve the accuracy of the modeling, sparse processing accuracy, but greatly reduced the amount of training data, thus effectively reducing the computational burden, so fast has been guaranteed.This article will improve the the LSSVM algorithm application in the field of population projections, the experiments show that the the prediction accuracy was proposed by algorithm of LSSVM...