Machine Learning for Glass Science and Engineering

October 17, 18, 19 from 11 a.m. to 2 p.m. EDT

Instructor: Mathieu Bauchy, University of California, Los Angeles (UCLA) 

This 3 day (9 hour) course will offer an introduction to machine learning applied to glass science & engineering and a hands-on tutorial.

Course description

Machine learning techniques are now ubiquitous in high-tech applications (e.g., search engine, face detection, spam identification, etc.) and allow computers to “learn” from existing data. More recently, machine learning methods have offered new paradigms to understand, engineer, and design glasses and materials in general. Machine learning offers a promising path to decode composition-property relationships in glasses, predict optimal glass compositions with tailored properties, pinpoint relevant structural patterns in atomistic simulations, and, more generally, guide and accelerate the design of new glasses.

This course will provide an introduction to machine learning and its application to glass science and engineering. Lectures will be complemented by a practical, hands-on tutorial on using machine learning for glass property prediction and glass optimization. Topics covered will include

• General introduction to supervised and unsupervised machine learning
• Review of existing machine learning methods and applications
• Overview of a complete machine learning pipeline: data collection, data cleaning, feature engineering, learning algorithm selection and training, hyperparameter optimization, testing and deployment
• Development of composition–property predictive models in glasses
• Optimization of glass formulations with optimal properties
• Integrations of machine learning, simulations, and experiments

Who should attend

This introductory course is targeted to students, scientists, or engineers who are interested in incorporating machine learning in their research or professional activities. No prior knowledge in machine learning or computer science is expected, although certain aspects of this course will also be relevant to individuals who are already familiar with machine learning. By the end of this class, participants are expected to:

• Understand the possibilities and limitations of machine learning,
• Be familiar with the different applications of machine learning in the field of glass science and engineering,
• Be able to choose the right machine learning method to solve a given problem,
• Have the necessary introductory theoretical background to understand previous studies focusing on machine learning and material informatics, and,
• Implement a machine learning model to predict glass properties as a function of their composition and prescribe optimal compositions featuring tailored properties.