Quantifying Disorder via Graph Analytics at NeurIPS 2023

Graph analytics provides a means to understand order-disorder transitions, which can critically impact the performance and lifecycles of materials in extremes. Using this approach, we can intelligently describe microstructural features and their changes in complex environments, moving beyond the limitations of humans to describe latent trends. This work is being presented by Arman Ter-Petrosyan at NeurIPS 2023.

From the abstract:

We present a method for the unsupervised segmentation of electron microscopy images, which are powerful descriptors of materials and chemical systems. Images are oversegmented into overlapping chips, and similarity graphs are generated from embeddings extracted from a domain‐pretrained convolutional neural network (CNN). The Louvain method for community detection is then applied to perform segmentation. The graph representation provides an intuitive way of presenting the relationship between chips and communities. We demonstrate our method to track irradiation‐induced amorphous fronts in thin films used for catalysis and electronics. This method has potential for "on‐the‐fly" segmentation to guide emerging automated electron microscopes.

Learn more about the Machine Learning for Machine Learning and the Physical Sciences Workshop.

Download the paper here.