Topological Data Analysis Based Machine Learning Models for Drug Design

Explore topological data analysis (TDA) based machine learning models for drug design in this 59-minute talk by Kelin Xia. Delve into a series of TDA-related models, including weighted persistent homology, persistent spectral models, and persistent Ricci curvature, and their integration with machine learning techniques. Discover how these persistent models characterize intrinsic multiscale information and provide molecular representations that balance data complexity and dimension reduction. Learn about generating molecular descriptors from various persistent attributes and their combination with machine learning models such as random forest, gradient boosting tree, and convolutional neural networks. Examine the extensive testing of these models on various databanks, particularly PDBbind datasets, and understand how persistent representations-based molecular descriptors significantly improve the performance of learning models in drug design compared to traditional molecular descriptors.