Machine learning has entered many scientific disciplines with great success. Algorithms and modeling tools have been developed and applied to many data processing tasks in physical sciences. Recently, it is noted it is important to design models and learning algorithms that can capture the underlying structure of the dataset. For example, symmetries inherent in the physical process are important and should be exploited to better model the data. Recent progress on multi-dimensional statistics and data dimension reduction also point to new directions in machine learning.

In this workshop, we bring together experts in physical sciences and mathematics to explore these topics. Through this workshop, we hope new interdisciplinary research direction can be identified and research teams can be formed. Furthermore, this will serve as an example of future collaboration between Mathematics and Physical Sciences