Sparse Band Selection for Support Vector Data Description Applications

Support Vector Data Description (SVDD) methods have been successfully applied to tasks such as hyperspectral anomaly detection [1] and spectral unmixing [2] [3]. Unfortunately, the performance of SVDD methods suffers when noisy or non-informative bands are present in the data. If a set of sparse bands could be identified for these techniques, the resulting data may improve SVDD performance while enjoying the benefits of decreased processing overhead. Although band selection has been investigated in previous efforts, this work builds on recent research that has resulted in the development of a theoretical framework for signal classification with sparse representation using L1 measures. This data-driven approach combines the classification power of the discriminative methods with the reconstruction property and a sparse representation that enables one to deal with signal corruptions: noise, missing data and outliers.