Title: Using AI to determine Space Groups from Powder Diffraction Using Over 1,000,000 Patterns

Neutrons and x-rays are excellent probes of the structures of materials. Modern instruments are generating increasing data volumes. With current technologies, structural classification from powder diffraction is manually intensive: one must precisely index peaks to get a reasonable starting point for space group determination, which is also labor intensive. In this talk, we explore the ability to use neural networks to classify the space group and Bravais lattices of materials based on their powder diffraction patterns. We extend upon previous work by increasing the data volume we use for model training and testing. We also carefully consider the effects of instrument resolution, duplicate entries in databases, radiation type, etc.