CIS Distinguished Seminar: Deep Learning, Differentiable Programming, and Software 2.0
Artificial neural networks have made spectacular progress recently in solving complex problems of pattern recognition, prediction and game playing. Among the paradigms that made this possible, deep learning is credited with the lion’s share, being even labeled a disruptive breakthrough in artificial intelligence by enthusiastic supporters. However, there exist also critical minds that voice concern about its apparent limitations, of which a black box operation that raises questions about the validity of deep learning as an alternative to formal methods, no matter what its current prowess may be. The talk will present current attempts to present deep learning from the manifold and functional programming perspectives, in order to lift the black box hindrance and open the door to formal neural network-based software development.
Mounir Boukadoum is an IEEE fellow and a full professor at The University of Quebec at Montreal (UQAM), Canada. He holds a PhD degree in Electrical Engineering degree from The University of Houston, Texas. He is one of the leading researchers in the applications of artificial intelligence and nature-inspired techniques to solve analysis and design problems, particularly in relation to biomedical outcomes, software engineering, robotics, signal processing, etc. Pr. Boukadoum has over 150 publications and is currently director of the Design and Fabrication of Microsystems research laboratory at UQAM (COFAMIC), president of the IEEE Computational Intelligence Society's chapter in Montreal and executive member of the Quebec Strategic Alliance for Microsystems research consortium (ReSMIQ). He is an active member of IEEE and member of three IEEE CAS conferences steering committees. He is also a cofounder of the IEEE NEWCAS conference and Co-Chair of the 2018 edition in Montreal. He has extensive industrial and academic experience on designing AI techniques to real-world applications including biomedical, software engineering, robotics, signal processing, etc.