Deep Dictionary Learning with An Intra-class Constraint

On the one hand, the dictionary is too large and the computational complexity In recent years, deep dictionary learning (DDL)has attracted a is too high for large-scale classification problems. On the great amount of attention due to its effectiveness for representation other hand, the original input training samples may contain learning and visual recognition. However, most existing noise, which leads to the inappropriate dictionary and suffers methods focus on unsupervised deep dictionary learning, failing from the problem of poor robustness. To address the abovementioned to further explore the category information. To make full issues, several supervised dictionary learning algorithms use of the category information of different samples, we propose such as D-KSVD [6] and LC-KSVD [7] have been a novel deep dictionary learning model with an intraclass proposed by introducing category information for dictionary constraint (DDLIC) for visual classification.