MAMMOth Catalogue

Explore the core modules of the MAI-BIAS and demonstrator.

This catalogue gathers the MAI-BIAS modules developed by the MAMMOth project. These hold research results and third-party libraries to perform various kinds of fairness assessment. Some will also guide you on how to apply mitigation strategies. The documentation found here is the same as the one shown by the respective tools. However, by organizing everything in one place, it becomes easier to understand all available options for the toolkit. Use the navigation links to overview datasets, models, and analysis methods.

Broadly, we offer three types of modules:

🗄 Dataset loaders

Load your own datasets or automatically use public datasets that are popular in fairness literature research.

🧮 Model loaders

Import a wide range of trained machine learning model formats or other artificial intelligence algorithms.

🔬 Analysis metrics

Includes traditional metrics quantifying disparities between sensitive groups and visualizations that aid explainability.

Visit our GitHub repository to quickly set up a demonstrator that lets you run all these modules.