author = "Filisbino, Tiene Andre and Giraldi, Antonio Giraldi and Thomaz, 
                         Carlos Eduardo",
          affiliation = "{National Laboratory for Scientific Computing} and {National 
                         Laboratory for Scientific Computing} and {Department of Electrical 
                         Engineering FEI}",
                title = "Ranking Methods for Tensor Components Analysis and their 
                         Application to Face Images",
            booktitle = "Proceedings...",
                 year = "2013",
               editor = "Boyer, Kim and Hirata, Nina and Nedel, Luciana and Silva, 
         organization = "Conference on Graphics, Patterns and Images, 26. (SIBGRAPI)",
            publisher = "IEEE Computer Society",
              address = "Los Alamitos",
             keywords = "Dimensionality Reduction, Tensor Subspace Learning, CSA, Face 
                         Image Analysis.",
             abstract = "Higher order tensors have been applied to model multidimensional 
                         image databases for subsequent tensor decomposition and 
                         dimensionality reduction. In this paper we address the problem of 
                         ranking tensor components in the context of the concurrent 
                         subspace analysis (CSA) technique following two distinct 
                         approaches: (a) Estimating the covariance structure of the 
                         database; (b) Computing discriminant weights through separating 
                         hyperplanes, to select the most discriminant CSA tensor 
                         components. The former follows a ranking method based on the 
                         covariance structure of each subspace in the CSA framework while 
                         the latter addresses the problem through the discriminant 
                         principal component analysis methodology. Both approaches are 
                         applied and compared in a gender classification task performed 
                         using the FEI face database. Our experimental results highlight 
                         the low dimensional data representation of both approaches, while 
                         allowing robust discriminant reconstruction and interpretation of 
                         the sample groups and high recognition rates.",
  conference-location = "Arequipa, Peru",
      conference-year = "Aug. 5-8, 2013",
             language = "en",
           targetfile = "Sibgrapi_2013.pdf",
        urlaccessdate = "2020, Dec. 05"