2019 Invited Speakers

 

Dr. Xiucai Ye, University of Tsukuba, Japan

 

 

Speech Title: Unsupervised Feature Selection by Data Structure Learning
Xiucai Ye and Tetsuya Sakurai
Department of Computer Science, University of Tsukuba, Japan
Center for Artificial Intelligence Research, University of Tsukuba, Japan

Abstract: In many applications such as machine learning and data mining, the data samples are often represented by a large number of features. The large number of features that often contain a lot of redundant and noisy information, make great challenges such as the curses of dimensionality and high computation cost. Feature selection is an efficient technique for data dimension reduction, which aims to extract the important features and eliminate the noisy ones. Unsupervised feature selection is much more difficult than supervised feature selection due to the lack of label information. In this talk, we will introduce a proposed framework for unsupervised feature selection which incorporates data structure learning and l2,1- norm sparse regularization. Statistical methods are utilized to learn the data structure. By imposing row sparsity on the transformation matrix, the resultant formulation optimizes for selecting the most discriminative features which can better capture both the global and local structure of data. Experimental results on different types of real-world data demonstrate the effectiveness of the proposed method.t Award (2016).

   
   

Copyright © 2019 11th International Conference on Machine Learning and Computing