Prof. Ferdinando Fioretto (postdoctoral researcher at the Georgia Institute of Technology)
Advances in artificial intelligence and data science have allowed the development of products that leverage individuals' data to provide valuable services. However, the use of this massive quantity of personal information raises fundamental privacy concerns. Differential Privacy (DP) has emerged as the de-facto standard to addresses the sensitivity of such information and can be used to release privacy-preserving datasets.
Despite its large theoretical value, when these private datasets are used as inputs to complex machine learning or optimization tasks, they may produce results that are fundamentally different from those obtained on the original data.
In this talk, I will review the notion of Differential Privacy and focus on the problem of releasing privacy-preserving data for complex data analysis tasks. I will introduce the notion of Constrained-Based Differential Privacy (CBDP) which allow us to cast the data release problem to an optimization problem whose goal is to preserve the salient features of the original dataset. Finally, I will discuss two applications of CBDP for large socio-technical systems related to the optimization of operations in transportation systems and energy networks.