RT DW A1 Patil, Shreyas Malakarjun . T1 Leveraging Sparsity in Deep Neural Networks for Training Efficiency, Interpretability and Generalization