Université de Toulouse
Deep learning: learning physics from data
Deep learning approaches are adapted to the context of technical data. We consider data generated by simulation tools and data generated by human activities. State of the art in model reduction methods are combining the data with modeling information, extracted from the simulation tool that generated the data. We will show that it is possible to extract modelling information from data. In other words, it is possible to learn physics from data without calling the simulation tool. Being non-intrusive, the proposed methods can be applied to any kind of data.
