Sparse Coding with Outliers

Sparse coding is invalid to learn parts-based representations when data is corrupted by outliers. In this paper, matrix completion is considered into sparse coding to handle outliers and a novel sparse coding method is proposed to learn a robust subspace. Experiments on the ORL dataset with salt and pepper noise and contiguous occlusion demonstrate that our proposed sparse method is more effective and robust in achieving a robust subspace.

[1]  John Shawe-Taylor,et al.  MahNMF: Manhattan Non-negative Matrix Factorization , 2012, ArXiv.

[2]  Yurii Nesterov,et al.  Smooth minimization of non-smooth functions , 2005, Math. Program..

[3]  Edmund Y. Lam,et al.  Non-negative matrix factorization for images with Laplacian noise , 2008, APCCAS 2008 - 2008 IEEE Asia Pacific Conference on Circuits and Systems.

[4]  Licheng Jiao,et al.  A fast tri-factorization method for low-rank matrix recovery and completion , 2013, Pattern Recognit..

[5]  Rajat Raina,et al.  Efficient sparse coding algorithms , 2006, NIPS.

[6]  David J. Kriegman,et al.  Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection , 1996, ECCV.

[7]  Emmanuel J. Candès,et al.  Exact Matrix Completion via Convex Optimization , 2008, Found. Comput. Math..

[8]  S T Roweis,et al.  Nonlinear dimensionality reduction by locally linear embedding. , 2000, Science.

[9]  H. Sebastian Seung,et al.  Learning the parts of objects by non-negative matrix factorization , 1999, Nature.

[10]  Yuxiao Hu,et al.  Face recognition using Laplacianfaces , 2005, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  V. Kshirsagar,et al.  Face recognition using Eigenfaces , 2011, 2011 3rd International Conference on Computer Research and Development.