Select Page


About the Program

The Mines Initiative for Data-Driven Materials Innovation (MIDDMI) is a collaboration between Colorado School of Mines and Citrine Informatics dedicated to educating Mines students on the fundamentals of materials informatics. During the MIDDMI launch phase of spring semester 2018, MIDDMI Fellows will work on materials design problems amenable to machine learning (ML) while developing best practices for data management and applying machine learning models to their research.

The four foundations of MIDDMI research projects are highlighted below:


The four foundations of MIDDMI research projects


Applying machine learning to materials problems requires new approaches to data management. MIDDMI Fellows will learn how to properly capture experimental and computational parameters so that their data are amenable to machine learning and also highly re-usable by other researchers. Current and subsequent datasets will adhere to the FAIR data principles, which means data will be Findable, Accessible, Interoperable, and Reusable. The FAIR Data Principles focus on enhancing data reusability and machine readability. Read more about the FAIR principles in this article.


Building from the foundation of proper data capture and storage, Fellows will learn how to structure data for input to machine learning models.


Once research data has been properly structured for machine learning, Fellows will learn how to interpret and build ML models on the Citrination platform. These models will be capable of guiding subsequent research, design, and discovery efforts.


MIDDMI Fellows will learn about a data-driven approach to materials design called sequential learning (SL). SL is an iterative process that will assist Fellows in rational exploration of high-dimensional materials design spaces.