Research On Semi-supervised Feature Selection Algorithm Based On Multi-view

In dealing with computer problems,the dimension of the real data is usually high.There are a lot of irrelevant features,which brought some difficulties to solve practical problems.In order to improve the accuracy,feature selection methods are proposed.These methods are very effective for dimensionality reduction,which can directly select the optimal feature subset from the original feature of the data.Therefore,the research on this topic has become a hot research topic in the field of machine data mining.In solving the practical problems,we can find that there are many perspectives in the data,and multi-view learning is also the key research topic in the process of machine learning.If we can find the hidden relationship between the complementary data,the effect of learning will be greatly enhanced.However,with the development of modern society,the large-scale application of data has increased the difficulty of data acquisition and marking.So,how to get the data from the perspective of the relationship between the multi-views,and select the maximum correlation with the smallest subset of redundancy,are the main contents of this paper.According to the latest developments of the calculation and the research field,this paper proposes a new semi-supervised feature selection algorithm based on multi-views,which can effectively obtain multi-views of the mutual information and relationship,and analyzes the various features of the redundancy between different perspectives.By combining several labeled data and unlabeled data,the feature selection and clustering learning are implemented together,which solves the unlabeled problem in multi-views learning.The main contributions of this paper are as follows:(1)In this paper,we construct an improved parallel SVM,taking a plurality of parallel computing data SVM classifier.This method ensures the promotion of performance and shortens training time.(2)We take the redundancy relation of each feature in each view into account in the multi-views feature selection processing.