How Computational Shortcuts&Hardware Choices Impact Fairness in AI

​Training AI systems can be expensive and there is always a demand for efficiency, both in hardware choices and model simplifications (e.g., network pruning and activation linearization). The issue is that these decisions affect fairness in machine learning systems.

​Our BuzzRobot guest, Nando Fioretto, an assistant professor at the University of Virginia, will discuss how computational techniques intended to reduce runtime and resource usage can inadvertently introduce fairness disparities in ML models.

​The talk will also analyze the impact of hardware choices on model generalization and fairness.

​Speaker bio:

​Nando Fioretto is an assistant professor at the University of Virginia. He works at the juncture of Machine Learning, privacy, optimization, and fairness. His research has been recognized with the 2022 Caspar Bowden PET award, the IJCAI-22 Early Career spotlight, and various other industry and research awards.