Division/Committee: [COMP: Division of Computers in Chemistry]

Deep generative models have seen widespread adoption in computational chemistry in recent years. These models are typically trained to generate new data points that follow the underlying data distribution. Popular deep generative models include variational autoencoders, normalizing flows, and diffusion models. They have been applied to a wide range of tasks, from accelerating enhanced sampling, generating small molecule structures and protein conformational ensembles, to discovery of new material and new-to-nature protein scaffolds. Beyond data generation, deep generative models also offer capabilities such as evaluating structure probability distributions and enhancing the accuracy of quantum mechanics calculations. This symposium aims to provide a platform for researchers to share cutting-edge research and discuss future directions for applying deep generative models to complex problems in computational chemistry, structural biology, and material sciences. We welcome participants from diverse fields, including academic and industry researchers working on drug discovery, biophysics, materials science, protein engineering, and electronic structure modeling. The symposium will be of interest to researchers and students engaged in computational chemistry and/or machine learning.

Ming Chen

Purdue University

Appearances