Our paper on low-rank approximation using elastic-net is accepted to CVPR 2015

[2015/03/09]

The following paper is accepted to IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2015):

  • Elastic-Net Regularization of Singular Values for Practical Subspace Learning by Eunwoo Kim, Minsik Lee, and Songhwai Oh
    • Abstract: Learning a low-dimensional structure plays an important role in computer vision. Recently, a new family of methods, such as l1 minimization and robust principal component analysis, has been proposed for low-rank matrix approximation problems and shown to be robust against outliers and missing data. But the methods often require a heavy computational load and can fail to find a solution when highly corrupted data are presented. In this paper, an elastic-net regularization based low-rank matrix factorization method for practical subspace learning is proposed. The proposed method finds a robust solution efficiently by enforcing a strong convex constraint to improve algorithm’s stability. It is shown that any stationary point of the proposed algorithm satisfies the Karush-Kuhn-Tucker (KKT) optimality conditions. The proposed method is applied to a number of low-rank matrix approximation problems to demonstrate its efficiency in the presence of heavy corruptions and to show its effectiveness and robustness compared to the existing methods.