The STUDIA UNIVERSITATIS BABEŞ-BOLYAI issue article summary

The summary of the selected article appears at the bottom of the page. In order to get back to the contents of the issue this article belongs to you have to access the link from the title. In order to see all the articles of the archive which have as author/co-author one of the authors mentioned below, you have to access the link from the author's name.

  
       
         
    STUDIA INFORMATICA - Issue no. Sp.Issue%201 / 2009  
         
  Article:   DECOMPOSITION METHODS FOR LABEL PROPAGATION.

Authors:  LEHEL CSATÓ, ZALÁN BODÓ. 		 
       
         
  Abstract:   In semi-supervised learning we exploit the "information" provided by an unlabelled data-set, in addition to the usually small training data-set. Acommonly used semi-supervised method is label propagation [9] where labels arepropagated from labelled to unlabelled data by employing similarity measures.The draw back of the algorithm is that its time requirement is prohibitive. This means that when a large amount of unlabelled data is used, a feasible algorithmis needed to compute the labels. In this paper we propose an approximation to label propagation. We divide the original problem into sub-problems that are computationally less prohibitive. A decomposition into K parallel sub-problems is considered where the sub-problems randomly and sparingly communicate with each other.

Key words and phrases. semi-supervised learning, kernel methods, label propagation. 		 
         
     
         
         
      Back to previous page