Eager Pruning: Algorithm and Architecture Support for Fast Training of Deep Neural Networks | IEEE Conference Publication | IEEE Xplore