The availability of large datasets has led to a paradigm shift in many fields of science and engineering and fueled an increasing use of data-driven methods. In fluid mechanics, there have been a growing interest in machine learning (ML) algorithms for modeling and optimization. The complex, and multiscale nature of fluid flows, however, poses new challenges to the available ML algorithms, which are not commonly developed for problems with such a wide range of spatiotemporal scales. On the other hand, the generation of big data, which is normality in typical applications of ML, is still a costly task – both experimentally and numerically – in fluid mechanics. Furthermore, issues such as ‘informing an ML algorithm about the known physics of problems’ and ‘extracting physical knowledge out of data’ call for special attention if ML is to be deployed to its full potential in this field.

Despite all challenges, in recent years ML has been used increasingly in a diverse range of fluid mechanics problems such as turbulence modeling, flow feature recognition, flow control, optimization, and uncertainty quantification. The proposed symposium intends to provide a forum for the exchange of ideas and experience in this growing field of research. The main goal is to address the above-mentioned challenges and move towards the use of data-driven methods as robust and reliable tools for modeling and optimization of fluid mechanics related problems. The symposium will also serve as an opportunity to initiate shared use of data and codes.