Supervised Convolutional Sparse Coding
This work addresses image representation for restoration tasks, but it is incremental as it extends an existing model with supervised regularization.
The paper tackled the problem of learning discriminative dictionaries in convolutional sparse coding by proposing a supervised approach, resulting in more semantically relevant filters and improved image reconstruction on unseen data.
Convolutional Sparse Coding (CSC) is a well-established image representation model especially suited for image restoration tasks. In this work, we extend the applicability of this model by proposing a supervised approach to convolutional sparse coding, which aims at learning discriminative dictionaries instead of purely reconstructive ones. We incorporate a supervised regularization term into the traditional unsupervised CSC objective to encourage the final dictionary elements to be discriminative. Experimental results show that using supervised convolutional learning results in two key advantages. First, we learn more semantically relevant filters in the dictionary and second, we achieve improved image reconstruction on unseen data.