Pruning Binarized Neural Networks: A Dedicated Framework and Globally Weighted Algorithms | ResearchPod