Preserving Deep Representations In One-Shot Pruning: A Hessian-Free Second-Order Optimization Framework

Open in new window