AI-MI–supported research is on the program at KDD 2026, the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining, running August 9–13 at ICC Jeju in Korea.
“From Noisy STEM to Crystal Structure: Evidence-Structure CoDiffusion under Composition Constraints,” by Guangyao Chen and Fengqi You of Cornell University, opens the conference’s AI-driven Scientific Discovery session in the AI for Sciences track on Thursday, August 13th.
The work goes after a practical bottleneck in materials characterization. Determining a crystal structure normally demands clean, high-quality data, and real microscopy is noisy. The paper introduces SCCD, an evidence-structure co-diffusion model that takes a single noisy scanning-transmission-electron-microscopy image together with a known chemical composition and returns a complete, simulation-ready crystal structure as a CIF file.
Read the paper: doi.org/10.1145/3770855.3818950