Promoting Fairness in AutoML Systems

About this project

Automated machine learning (AutoML) was proposed as an artificial intelligence-based solution for automating the tasks of applying machine learning to real-world problems. In this context, several algorithmic challenges emerge in a variety of ways, including societal bias embedded in training datasets, decisions made during the development of an ML system, and through complex evaluation loops that arise when an ML system is deployed in the real world. To address this challenge, this project aims to develop a dynamic strategy to decide when and on which ML model to conduct unfairness mitigation according to accuracy/fairness trade-offs.