Detecting Discriminatory Outcomes in Classification Models
Lending, policing, and hiring are some of the many areas where Machine Learning can harm disproportionately the most vulnerable groups in our society. This can lead to discrimination and long-lasting negative impact in society. It is therefore crucial to understand unfair treatment in AI to prevent automated discrimination at scale. The fundamental techniques to analyze and detect bias in Machine Learning decision can be explained through simple metrics applied to model outcomes. The aim of this presentation is to pass this knowledge to empower ML practitioners to challenge how Machine Learning is implemented.
Valeria is a Data Scientist at Lloyds Banking Group, specializing in the design and development of scalable Machine Learning solutions for different business areas of LBG and their customers. Her current work at LBG focuses on building tools and processes to detect and mitigate bias in ML models. Before joining LBG, Valeria started her career in Cambridge researching on the economics of privacy at Microsoft Research and working for TAB, a Fintech startup. Valeria is a strong advocate of ethics and responsibility in AI as well as bringing more diversity into tech teams.