A novel automatic grouping algorithm for feature selection

作者：Yuan Qiulong; Fang Yuchun^*

来源：2nd Chinese Conference on Computer Vision, CCCV 2017, 2017-10-11 To 2017-10-14.

DOI：10.1007/978-981-10-7305-2_50

• Application area
• Automatic grouping
• Feature selection methods
• Grouping algorithm
• High dimensional feature
• Important features
• Intrinsic data structure
• Mutual informations

Feature selection is used in many application areas relevant to expert and intelligent system such machine learning, bioinformatics and image processing. Feature selection plays an important role in reducing the dimensionality of high-dimensional features. However, traditional feature selection methods are not able to intelligently learn intrinsic data structures. In this paper, we proposed a novel feature selection method, which can automatically learn grouping structure relation among features. Experiments are conducted on the selection of both raw features and statistically handled features. Experimental results demonstrate that the proposed method can identify important features by automatic grouping, and outperforms the other methods on several public data sets. Moreover, by using parallel computing, the training time consumed by our method is only 50% of that of the traditional methods.

• 出版日期2017
• 单位上海大学