A Faster K-SVD

Omeiza Olumoye, Mandela Bright Quashie

Keywords: K-SVD, PCA, Dictionary Learning, Sparse Coding, Norm

Abstract: This work is about improving the performance of the K-SVD, which was proposed by Aharon et al., 2006. The K-SVD is considered to be a state-of-the-art algorithm. In this work, we propose a way to make the algorithm better. Specifically, making it faster without compromising the quality of the images We achieved this by replacing the stage where the best rank-1 approximation using singular value decomposition (SVD) is updated in the algorithm with an l1-norm principal component analysis (l1-norm PCA).

Submission Category: Machine learning algorithms

Submission Number: 5