Department of Chemistry - PhD Student Seminar - Machine Learning Empowers Organic Chemistry - Frontier Methods and Case Studies

Student: Mr. Yizhou WANG

Department: Department of Chemistry, HKUST

Supervisor: Prof. Haibin SU

Co-supervisor: Prof. Yangjian QUAN

Abstract

Traditional trial-and-error organic synthesis suffers from low efficiency and high resource costs when confronting some novel or unknown chemical reactions. Machine learning (ML) combined with quantum chemical calculations such as density functional theory (DFT) establishes a data-driven research paradigm to address these bottlenecks. In this seminar, we would like to systematically introduce some core cheminformatics techniques, including multi-type molecular descriptors, several dimensionality reduction methods like PCA, and the application of regression models, which convert molecular structures into quantifiable fingerprints for property prediction. The review summarizes two mainstream data acquisition paths (experimental and DFT-computed descriptors) and highlights unified advantages of ML: shortening R&D cycles, reducing trial waste and uncovering reaction mechanisms. Besides, we will introduce several case studies in different problems of organic chemistry to better understand the wide application of machine learning in organic chemistry. Finally, future directions covering active learning robotic platforms, multi-scale generative molecular models and chemistry-specific large language models are outlined, demonstrating ML as an indispensable engine for intelligent organic chemistry innovation.