Implementation of Learning to Protect Communications with Adversarial Neural Cryptography in PyTorch
Caution: Details of implementation differ much from the original paper. I tried to use dense(fc) layers with relu activation, instead of dense+conv1d+sigmoid. See model/model.py
.
pip install -r requirements.txt
(PyTorch, Numpy, TensorboardX)
python trainer.py -c config/default.yaml -n [name of run]
- You may copy
cp config/default.yaml config/config.yaml
and change parameters (e.g. size of plain/key/cipher, depth of NN, …) to experiment with your own setting.
tensorboard --logdir logs/
Result with config/default.yaml
, trained on GTX 1080 for 1 hour.
- Accuracy: (Green: Bob, Gray: Eve, Orange: Random guess)
- Loss: (Green: B, Gray: E, Magenta: AB)
Apache License v2.0