Aula aberta no âmbito da Unidade Curricular de Aprendizagem Profunda, por Hugo Penedones (Inductiva).

Fundamental Sciences and Engineering have been using numerical simulation methods for decades, accelerating the discovery of knowledge and the development of new technologies. At the same time, the AI and ML communities were busy developing methods that replicate human capabilities of perception, pattern recognition and reasoning. In recent years, we are witnessing a confluence of the fields: Machine Learning is being used, in many different ways, to expand the boundaries of what Scientific Computing can do. In this talk we will give an overview of current trends and explain what Graph Neural Networks and Physics-Informed Neural Networks are.

Bio: Hugo Penedones has a degree in Informatics and Computing Engineering from the Univ. of Porto (PT) and is a Machine Learning research engineer who has worked in computer vision, search, bioinformatics and reinforcement learning, at places such as the Idiap Research Institute and EPFL (both in Switzerland), Microsoft and Google DeepMind (London, UK). He has a particular interest in the applications of Deep Learning to fundamental sciences and engineering and he is the co-founder and CTO at Inductiva Research Labs.