Presentation - Title Trustworthy AI: The role of the Hardware

Abstract

In the recent years, AI algorithms became so accurate that started to surpass human being in a wide range of tasks and are now currently used in applications that would be considered science-fiction only few years (months) ago. As AI advances across economies and societies, stakeholder communities are actively exploring how best to encourage the design, development, deployment and use of AI that is human-centered and trustworthy. Trustworthiness is a wide multi-disciplinary research field including mathematics and computer science for which the goal is to evaluate Fairness, Transparency, Explainability and Robustness of a given AI model knowing the data used for training and validation. However, one of the neglected aspects is the role of the hardware used to deploy the final application. Indeed, hardware is subject to external perturbations that can jeopardize the execution of AI workloads leading to misclassification and eventually to dramatic consequences when AI is used in safety-critical applications, such as autonomous driving. This talk will present the main hardware threats and discusses the main consequences on the choice of a given hardware architecture to achieve by design Trustworthiness. 

Biography - Dr. Alberto Bosio