Alternating Direction Method of Multipliers for Solving Dictionary Learning Models

In recent years, there has been a growing usage of sparse representations in signal processing. This paper revisits the K-SVD, an algorithm for designing overcomplete dictionaries for sparse and redundant representations. We present a new approach to solve dictionary learning models by combining the alternating direction method of multipliers and the orthogonal matching pursuit. The experimental results show that our approach can reliably obtain better learned dictionary elements and outperform other algorithms.