Over the last few decades sparsity has become a driving force in the development of new and better algorithms in signal and image processing. In the context of the late deep learning zenith, a recent pivotal work showed that deep neural networks can be interpreted and analyzed as pursuit algorithms seeking for sparse representations of signals belonging to a multilayer synthesis sparse model. In this talk, I will review recent contributions showing that this observation is correct but incomplete, in the sense that such a model provides a symbiotic mixture of coupled synthesis and analysis sparse priors. We will make this observation precise and use it to expand on uniqueness guarantees and stability bounds for the pursuit of multilayer sparse representations. We will then explore a convex relaxation of the resulting pursuit and derive efficient optimization algorithms to approximate its solution. Importantly, we will deploy these algorithms in a supervised learning formulation that generalizes feed-forward convolutional neural networks into recurrent ones, improving their performance without increasing the number of parameters the model.
Back to Workshop IV: Deep Geometric Learning of Big Data and Applications