Skip to content

Create certifiably secure scikit-learn compatible machine learning pipelines. Based on "Certified Defenses for Data Poisoning Attacks" by Steinhardt et al.

Notifications You must be signed in to change notification settings

adversarialml/certml

Repository files navigation

Build Coverage Maintainability Requirements

CertML

NOTE: Under development and presently unstable.

CertML allows for the creation of scikit-learn compatible machine learning pipelines that are certifiably secure based on Certified Defenses for Data Poisoning Attacks by Steinhardt et al.

About

Create certifiably secure scikit-learn compatible machine learning pipelines. Based on "Certified Defenses for Data Poisoning Attacks" by Steinhardt et al.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages