This code replicates the experiments in the paper "Defending Evasion Attacks via Adversarially Adaptive Training", published at IEEE Big Data 2022 (paper)
In this project, we use python 3.7.0 and dependencies:
- Install python 3.7.0: https://www.python.org/downloads/release/python-370/
- Install pip for python: https://pip.pypa.io/en/stable/installation/#
- Install dependencies:
pip install -r pip-requirement.txt
Note: In order to avoid conflicts between the dependencies from other projects, we highly recommend using python virtual environment or Anaconda. The details can be found here: https://docs.python.org/3/tutorial/venv.html
To work on our source code, you might want to modify the following files:
models.pyModels that we used in AAD experiments.aad_cnn.pyFunctions to train adversarially adaptive model with MNIST dataset.aad_nn.pyFunctions to train adversarially adaptive model with COMPAS dataset.
To run the code aad_cnn.py and aad_nn.py, use terminal with these arguments:
`--lmd` Lambda value
`--gamma` Gamma value
`--seed` random seed
`--batch_size` batch size
`--meta_batch` number of meta batchs
`--lr` learning rate
`--gpu_id` gpu id for pytorch to use if available
`--epochs` number of training steps
`--model_name` name of the model to be saved
data_processor.pyClasses and functions to build training and test data that used in our scenarios.LR_baseline.py,CNN_pre_baseline.pyandCNN_dec_baseline.pyLR and CNN baselines in our experiments.test_white_box.pyTest AAD model on white box scenario.
Due to file size limitation, we only include the source code for this project. The input data for this repo can be generated following the instructions in "Reproducibility_supplementary"
The full source code + data can be downloaded at downloaded at https://tinyurl.com/yxt5e869