Artificial Intelligence Data Fairness and Bias

The first module, Fairness and protections in machine learning, introduces essential concepts related to fairness and protections. It explores model parity, protecting groups and individuals, and the challenges of imperfect modeling. Participants will gain insights into the equality conundrum, COMPAS system, and the importance of fairness in various real-world scenarios. The module incorporates knowledge checks and quizzes to reinforce learning.

The second module, Building fair models: theory and practice, delves into the theoretical and practical aspects of building fair models. It covers algorithms inside algorithms, testing fair loan decisions, deploying fairness in practice, and adversarial models. Participants will examine unfairness visualization, research papers on debiasing, and the pragmatic approach to getting to fair models. The module includes knowledge checks and quizzes to assess understanding.

In the third module, Human factors: minimizing bias in data, the focus shifts to human factors influencing bias in data. Participants will explore bias awareness, building exploratory training sets, and the role of game theory in minimizing bias. The module also provides an in-depth understanding of cognitive biases and offers practical exercises such as Monster Match. Knowledge checks and quizzes are incorporated to consolidate learning.