PLOS_Flexible_Networks

¹ Equal contribution

Instructions:

The code for implementing the flexible layer is found in network_definitions.py.
To recreate main results, run train_VGG16.ipynb, import the network of interest from network_definitions.py. This script saves validation accuracies. Uncomment the line to save model weights if you wish to run further analyses.
Once you have trained a VGG16 model, run adverserial_FGSM.ipynb (to recreate Fig. 4D) or adverserial_PGD.ipynb to recreate the adverserial attack results.
Train a network on a downsampled subset of data using this script: train_downsampled.ipynb. To recreate our results run for seeds 1-10.
The spectral analysis figures are created with test_robustness.ipynb, out_of_distribution_inference.ipybn, or bandpass_inference.ipynb.

Provide feedback

Name		Name	Last commit message	Last commit date
Latest commit History 12 Commits
README.md		README.md
adverserial_FGSM.ipynb		adverserial_FGSM.ipynb
adverserial_PGD.ipynb		adverserial_PGD.ipynb
bandpass_inference.ipynb		bandpass_inference.ipynb
network_definitions.py		network_definitions.py
out_of_distribution_inference.ipynb		out_of_distribution_inference.ipynb
requirements.txt		requirements.txt
test_robustness.ipynb		test_robustness.ipynb
train_VGG16.ipynb		train_VGG16.ipynb
train_downsampled.ipynb		train_downsampled.ipynb