Jinglai Shen, Department of Mathematics and Statistics.
Saeed Damadi, Department of Mathematics and Statistics.
Compressing deep neural networks aims at finding a sparse network that performs as well as a dense network but with significantly less parameters. It is also called pruning. As a result of pruning, the energy consumption reduces, hardware requirements are relaxed, and responses to queries become faster. The pruning problem yields a constrained, stochastic, nonconvex, and non-differentiable optimization problem with a very large size. Efficient optimization schemes will be developed.