Corpus ID: 234358920

Towards improving discriminative reconstruction via simultaneous dense and sparse coding

@inproceedings{Tasissa2020TowardsID,
  title={Towards improving discriminative reconstruction via simultaneous dense and sparse coding},
  author={Abiy Tasissa and E. Theodosis and Bahareh Tolooshams and Demba E. Ba},
  year={2020}
}
Discriminative features extracted from the sparse coding model have been shown to perform well for classification and reconstruction. Recent deep learning architectures have further improved reconstruction in inverse problems by considering new dense priors learned from data. We propose a novel dense and sparse coding model that integrates both representation capability and discriminative features. The model considers the problem of recovering a dense vector $\mathbf{x}$ and a sparse vector… Expand
On the convergence of group-sparse autoencoders