Introduction

Machine Learning in Chemistry

In this course Professor Mark Tuckerman (New York University) introduces us to the concept of machine learning, a relatively new approach in computer science to solving complex problems. Machine learning has already proved to have widespread applications in industry and science, but in this lecture series Professor Tuckerman the applications of machine learning to problems in the field of chemistry. We begin by (i) introducing the term ‘machine learning’, everyday uses in computing and in the history of science, as well as its origin; (ii) followed by discussing where machine learning can be applied in chemistry, the problems that face the chemical industry as well as its academia; (iii) then moving to discuss molecular dynamics, the physics that govern molecules, and the ways in which machine learning can save hugely on computing costs in these areas; (iv) and finally investigating more closely and specifically the progress of modern research optimising machine learning for purposes in chemistry.

Introduction

In this first mini-lecture we start by introducing machine learning. Where do we come across machine learning in our everyday lives, and where has it been seen in the history of science? Most shopping portals online employ machine learning practices to learn better what their customers might buy, and also use it to suggest similar products to customers to encourage further purchases. Through this we begin to understand how machine learning models can be trained: how we teach them to understand what we want to see as outputs. We then move onto the inspirations of machine learning and neural networks, the human brain. What similarities do they share? To finish off, we learn about how machine learning advanced the field of protein folding hugely in 2020, and the exciting prospects of this for future research. Finally we look at an example of a machine learning model: the kernel method.