Using Machine Learning For Quantum Chemistry

Introduction:

• This is the README.md file for the project "Using Machine Learning For Quantum Chemistry", produced by Reza Sadoughian zadeh, @sezar543 and Peter Gysbers.
• This project was done for the course CPSC 340 at UBC during the fall semester in 2019 .

Objective:

The goal of this project was to review the following paper and discuss future research that can be performed to improve results. link

Preface:

We present several different machine learning methods which predict the chemical properties of molecules based on their configurations. We discuss gradient domain machine learning (GDML) which is a kernel regression technique to learn the force-fields which cause molecular dynamics. In addition "SchNet" is a neural network architecture which extends convolutional networks. It has various extensions which improve it.

As machine learning has developed, finding new applications in image recognition, language processing and recommendation systems, applications of machine learning to quantum chemistry have also progressed. This development has advanced through linear regressions, to kernel regressions, to neural networks and most recently to advanced tensor networks with connectivity and activation functions that are highly customized to this particular prediction problem.