This coming Spring MSE will offer MSE403: Data Sciences and Analytics for Materials Engineers, as a 3rd and 4th year elective, taught by Prof. Jason Hattrick-Simpers.
This course, introduces the elements of data sciences, materials informatics and data analytics in materials science and engineering. The focus will be on the applications of this emerging field for accelerated materials development. Guest lectures from expert practitioners in the field of AI in materials science will provide “bleeding edge” examples of autonomous materials systems, on the fly data acquisition, and the integration of theory and machine learning.
What will students learn from this course?
The students will learn the fundamentals of machine learning in materials science and by the end of the semester they will be able to:
- Formulate materials science problems and select a machine learning model from the supervised/unsupervised, classification/regression, and interpretable/explainable archetypes.
- Create meaningful and descriptive features of materials to maximize the predictive performance of a selected model.
- Critically evaluate the predictions of a ML model for over-fitting, range of applicability, uncertainty, and physical plausibility.
What sort of machine learning approaches will the students be expose to during this course?
The students will also be exposed to machine learning approaches such as supervised and unsupervised learning; linear, non-linear, and logistic regression, decision trees, and artificial neural networks. They will also be trained on programming these algorithms in python and applying them for a set of case studies pertaining to structure-property relations in materials science, alloy design, additive manufacturing, and green energy technologies.
What are the prerequisites for this course?
Previous experience in coding is not a requirement for the class, we will teach the appropriate coding skills (and point out resources) as they come up. However, those with a background in python and scikit-learn are encouraged to register.
What type of homework/projects will be involved?
Homework and projects will focus on building models from real materials data sets (experimental and theoretical) and using those models to predict new materials with potentially interesting properties.