AI and machine learning (ML) offer important new ways of solving today’s complex engineering problems. One of the major active innovation in this area is the combination of machine learning (ML) and physics-based models (PM). As a scientist in computational mechanics, I work on making mathematical, numerical and data driven models to describe the physics of the world around us. A well-made PM should enable us to simulate complex phenomena and accurately predict future responses. Such models have already been successfully applied in many engineering fields including mechanics, materials, aerospace, medicine and others. However ability of ensembled/hybrid ML and PM models to simulate and predict is of central importance where there is a lack of direct process based theoretical knowledge about the system and bear a very promising and exciting prospect. This presentation showecases my previous work of simulation with PM and its transition to ML approach as well as my recent work combining ML with PM. In addition, many questions are raised about ML modeling to weigh their respective benefits, drawbacks and tradeoffs; and to make what are often difficult choices for selecting the right model.

Dr. Abderrachid Hamrani was a postdoc researcher at McGill university (Canada, 2018-2020). Before that, he was an assistant professor at University of Boumerdes (Algeria, 2011-2018). He received his Ph.D. in Mechanical engineering (manufacturing processes) from Arts et Métiers Institute of Technology ParisTech (France, in 2016). His current research interests include AI and machine learning, advanced modeling and simulation in computational mechanics (solids and fluids), optimization techniques and inverse analysis. He has published over 22 scientific articles, which includes 12 peer-reviewed journal articles, 1 chapter-book.