Li Jial1, Su Ji1, Lu Jion1
1National University of Singapore, Chemistry Department
In the rapidly evolving field of materials science, Scanning Tunneling Microscopy (STM) and Atomic Force Microscopy (AFM) generate vast amounts of data that require advanced computational techniques for effective analysis, including automation, interpretation, and prediction. This tutorial will introduce the latest computational algorithms designed to process and analyze microscopy data. We will start with image-tailored algorithms such as convolutional neural networks (CNNs), which are adept at handling continuous image data to extract meaningful features. Following this, we will explore graph neural networks (GNNs), which are suited for analyzing non-Euclidean graph data, such as molecules and crystals, enabling the extraction of atomic and bonding information at the microscopic level. This session will include practical demonstrations and hands-on examples to illustrate the implementation and benefits of these algorithms in real-world microscopy applications. By the end of the tutorial, attendees will gain a solid understanding of how to apply CNNs and GNNs (e.g., Chemprop) to enhance their analysis of STM and AFM data, ultimately advancing their research in materials science. This tutorial is intended for both novice and intermediate researchers who aim to leverage machine learning techniques in their microscopy studies.