MSc. Data Science for Public Policy

Deep Learning
Bias in AI: detection and mitigation

Authors:

• Jorge Roa, 226454@students.hertie-school.org
• Carlo Greß, 216319@students.hertie-school.org
• Hannah Schweren, 216579@students.hertie-school.org

This notebook provides our code and explanations to enable users to detect & mitigate bias, using the example of the COMPAS Recidivism Risk Score Data and Analysis Dataset. Users will be equiped with concrete strategies to first detect, and secondly mitigate bias. Following this tutorial is a first step to become awareness of biases in Algorithms and promote ethical AI - especially in the intersection with policy decisions.