roby (ROBustness analYzer) is a Python tool to perform robustness evaluation of neural network classifiers. Given a trained model, a classified dataset and the list of classes, alterations are applied and the robustness is computed based on the accuracy threshold defined.

• How to use roby
• Tutorials and application scenarios
• How to extend roby
• Examples of use

A full documentation of the roby's APIs can be found at https://fmselab.github.io/roby/apis/