Palestras e Seminários
13/09/2022
14:00
Auditório Luiz Antonio Fávaro
Palestrante: Romain Giot
Responsável: André Carlos Ponce de Leon Ferreira de Carvalho (Este endereço de email está sendo protegido de spambots. Você precisa do JavaScript ativado para vê-lo.)
Deep Learning is omnipresent both in academic research and industrial applications. This is totally understandable as such solutions usually outperform standard methods. However such performance is often at the cost of a lack of understanding of the whole pipeline. These systems are seen as black boxes.
In this presentation, we will see how eXplainable Deep Learning can help to open such black boxes. We will first see the limitations of deep learning systems, what interpretable systems are, what post-hoc analysis is. The talk ends with the presentation of generalities of Visual Analytics systems related to eXplainable Deep Learning.
Romain Giot received his Ph.D. degree in biometric authentication at the University of Caen in 2012 and is an associate professor at the University of Bordeaux in a big-data visualization team since 2013. His researches are dedicated to visualization, machine learning and their junction in eXplainable AI (XAI). He co-authored dozens of peer-reviewed papers and is involved in several Program Committees.
