Compressive Sensing Classifier Based on K-SVD

In recent years, compressive sensing has attracted considerable attention in many fields including medical diagnosis. However, due to the small number of medical samples, traditional methods might not be able to classify the medical data very well. In this paper, we propose a novel classification model based on K-SVD named Cluster-KSVD, which can obtain orthogonal dictionary from over-complete dictionary. With the reduced dictionary, we apply the sparsified features for classification, such as medical diagnosis. The sufficient experimental results demonstrate that our model has superior classification performance than the state-of-the-art classifiers.