Supervised Deep Sparse Coding Networks

In this paper, we propose a novel multilayer sparse coding network capable of efficiently adapting its own regularization parameters to a given dataset. The network is trained end-to-end with a supervised task-driven learning algorithm via error backpropagation. During training, the network learns both the dictionaries and the regularization parameters of each sparse coding layer so that the reconstructive dictionaries are smoothly transformed into increasingly discriminative representations. We also incorporate a new weighted sparse coding scheme into our sparse recovery procedure, offering the system more flexibility to adjust sparsity levels. Furthermore, we have devised a sparse coding layer utilizing a ’skinny’ dictionary. Integral to computational efficiency, these skinny dictionaries compress the high dimensional sparse codes into lower dimensional structures. The adaptivity and discriminability of our 13-layer sparse coding network are demonstrated on four benchmark datasets, namely Cifar-10, Cifar-100, SVHN and MNIST, most of which are considered difficult for sparse coding models. Experimental results show that our architecture overwhelmingly outperforms traditional onelayer sparse coding architectures while using much fewer parameters. Moreover, our multilayer architecture fuses the benefits of depth with sparse coding’s characteristic ability to operate on smaller datasets. In such data-constrained scenarios, we demonstrate our technique can overcome the limitations of deep neural networks by exceeding the state of the art in accuracy.