This repository contains a PyTorch implementation of TRADI: Tracking deep neural network weight distributions . For now, there are just the code scripts for MNIST, we will try to add the Camvid codes. The codes for MNIST are on the folder MNIST_TASK.

To launch the codes please go to the MNIST_TASK folder and then type:

python main_train_TRADI.py

Results

We can the results of the Maximum class probability for a TRADI DNN.