Skip to content

LebronX/Neuro-Symbolic-Verification

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

The supplementary material for IJCAI submission of Neuro-Symbolic Verification of Deep Neural Networks.

To run the code, please first install Marabou (https://github.com/NeuralNetworkVerification/Marabou) with Python version 3.6.0. Then you should use pytorch to train a network with two sets of input, which like self.fc_input = nn.Linear(14 * 14, 12) and self.fc_input_prop = nn.Linear(14 * 14, 5) and train with the network with corresponding property you would like to verify. Of course, you could also use the network that we provide in networks folder.

The experiment results are also provided in the validate folder.

Basic command:

Network training: python3 mnist_equivalence_training.py --epochs 1 --save-model --seed 1

Verification: python3 run-test.py --network 1

About

Neuro-Symbolic Verification of Deep Neural Networks

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages