Periodic Graph Representation Learning and Transformers for Crystal Material Property Prediction
AIMS, National Institute of Standards and Technology, July 2023
Audience: Researchers, industry professionals, and government officials in materials science and AI.
Overview: Invited by NIST to present cutting-edge AI methods for materials science. I delivered a talk covering topics including the importance of periodic invariance, the way to better represent materials using graphs and self-connecting edges, the SOTA method Matformer proposed by us.
AI for Materials Geometric Representation Learning and High Tensor Order Property Predictions
AIMS, National Institute of Standards and Technology, July 2024
Audience: Researchers, industry professionals, and government officials in materials science and AI.
Overview: Invited the second time by NIST to present novel AI methods for materials representation learning and property prediction. I delivered a talk covering topics including how to achieve geometrically complete representations for materials, how to naturally inject physical constraints including symmetries and space groups into the network design, and the SOTA methods ComFormer and GMTNet proposed by us.
AI for Scientific Discovery of Proteins and Materials
D.E. Shaw Research, April 2024
Audience: Researchers and industry professionals in proteins science, materials science and AI.
Overview: Selected as the D.E. Shaw Research Doctoral & Postdoctoral Fellow. Presented cutting edge AI methods including ComFormer for materials discovery and LatentDiff for protein discovery.
To be updated.