MIT CINCS / Hamilton Institute Seminar

Wednesday, November 17, 2021 - 16:00 to 17:00
Passcode: mh-seminar

MIT CINCS (Communications Information Networks Circuits and Signals) / Hamilton Institute Seminar

Speaker: Professor Tess Smidt, Massachusetts Institute of Technology

Title: "Neural Networks with Euclidean Symmetry for Physical Sciences"

Abstract: To use machine learning to understand and engineer complex physical systems (e.g. materials for energy and computation and molecules and proteins for medicines), we need methods built to handle the “data types” of physical systems: 3D geometry and the geometric tensors. These are traditionally challenging data types to use for machine learning because coordinates and coordinate systems are sensitive to the symmetries of 3D space: 3D rotations, translations, and inversion. In this talk, I present Euclidean neural networks which naturally handle these data types. These networks eliminate the need for data augmentation -- the 500-fold increase in brute-force training necessary for a model to learn 3D patterns in arbitrary orientations. They are extremely data-efficient; they result in more accurate models and require less training data to do so. With these networks we are able to scale expensive quantum mechanical computer simulations to unprecedented system sizes and invent algorithms that can guide the design of atomic systems. I'll demonstrate some unique properties of Euclidean neural networks and show recent applications to increasing the accuracy and speed of molecular dynamics, predicting vibrational properties of crystals, and beyond.

Bio: Tess Smidt is an Assistant Professor of Electrical Engineering and Computer Science at MIT. Tess earned her SB in Physics from MIT in 2012 and her PhD in Physics from the University of California, Berkeley in 2018. Her research focuses on machine learning that incorporates physical and geometric constraints, with applications to materials design. Prior to joining the MIT EECS faculty, she was the 2018 Alvarez Postdoctoral Fellow in Computing Sciences at Lawrence Berkeley National Laboratory and a Software Engineering Intern on the Google Accelerated Sciences team where she developed Euclidean symmetry equivariant neural networks which naturally handle 3D geometry and geometric tensor data.