Description:

A2PM is a gray-box method for the generation of realistic adversarial examples. It benefits from a modular architecture to assign an independent sequence of adaptative perturbation patterns to each class, which analyze specific feature subsets to create valid and coherent data perturbations.

This method was developed to address the diverse constraints of domains with tabular data, such as cybersecurity. It can be advantageous for adversarial attacks against machine learning classifiers, as well as for adversarial training strategies. This Python 3 implementation provides out-of-the-box compatibility with the Scikit-Learn library.

License

For personal and professional use. You cannot resell or redistribute these repositories in their original state.

Share

• Share on Facebook
• Share on Twitter