Tools, paradigms, and methodologies from data science are increasingly being adopted in diverse areas of ChemE research and practice for their power in parsing, visualizing, extracting understanding, and guiding inquiry in the analysis of computational, experimental, and industrial data sets. In some cases these tools may be applied "off-the-shelf" with little or no modification, but in many cases their true value cannot be realized without adapting these methods with domain specific knowledge, or – more rarely – the development of entirely new tools tailored to particular data sets and applications. The focus of this session is to report and discuss methodological advances and new tools applied to any area in ChemE, including, but not limited to, process control, reaction engineering, thermodynamics, separations, transport, and energy and fuels. Innovations and advances in tools including, but not limited to, machine learning, QSPR, data-driven design, informatics, deep learning, virtual screening, data-driven control, mechanism inference, visualization, explainable artificial intelligence, regularization, incorporation and enforcement of physical laws and symmetries in artificial intelligence, active learning, and techniques for sparse data are of interest. The focus on this session will be specifically developing or tailoring data science methods as opposed to the application of existing methods.

Applications of Data Science to Molecules and Materials

Catalysis and Reaction Engineering Division (20), Computational Molecular Science and Engineering Forum (21), Information Management and Intelligent Systems (10E)