Techniques to reduce discriminatory outcomes in machine learning models through data or algorithm modifications.