Implementation of Semi-Supervised and Semi-Weakly Supervised Teacher/Student Learning approaches, from "Semi-supervised training of deep convolutional neural networks with heterogeneous data and few local annotations: an experiment on prostate histopathology image classification"