Symposium MT03—Automated and Data-Driven Approaches to Materials Development—Bridging the Gap Between Theory and Industry
Process development cycles of materials from discovery through synthesis towards manufacturing and devices is inherently slow (15-25 years). Today, there is parallel development of automation and robotics, high performance computing (HPC) and ubiquitous machine learning algorithms, the confluence of which, can better inform experiments and theory. This unique opportunity can potentially transform the way in which we do materials research, with faster learning, new tools and deeper insight into materials science problem. This can ultimately accelerate innovation and result in faster returns-on-investment for funding agencies and practitioners alike. Application of machine learning to materials science problems is still a nascent field, and many open questions remain. Especially intriguing is the unique nature of the material science as a testbed to test newly developed algorithms in machine learning – sparse datasets, varying qualities of labeled data and natural language processing to mine data from large bodies of literature are key features that can benefit both fields. In addition, new experimental capabilities can be built that can leverage upon the extended toolkit that machine learning and automation along with HPC can provide.
The purpose of this symposium is to convene experts in various domains with a shared interest in accelerating the rate of novel materials development. Specifically, this symposium will focus on bridging gaps between theory and industrial application, with an emphasis on the materials down-selection, accelerated testing, and industry & technology transfer.