Create a new virtual environment.
Install Jupyter Notebook.
Install the following required libraries, or use the environment.yml file to install the dependencies:
	conda install -c anaconda numpy
	conda install -c pytorch pytorch
	conda install -c pytorch torchvision
	conda install -c conda-forge matplotlib
	conda install -c anaconda pandas
	conda install -c anaconda scipy
Create a folder named 'files_npys' inside the current working directory.

1. Train the network on CIFAR-10.
2. Using the network, save the intermediate outputs for training set and test set.
   Beware: This step requires a high usage of RAM.
3. On training set, calculate the means and distances of samples to those means.
4. Apply normalization and softmax.
5. Perform thresholding on training set to choose the best threshold values.
6. On test set, calculate distances of samples to the means calculated on training set.
7. Apply normalization and softmax.
8. Perform thresholding on test set using the threshold values chosen on training set.
