4 Machine Learning in High-Energy Physics

Computers are useless. They can only give you answers.

– Pablo Picasso

Machine learning is an interdisciplinary field that deals with the general problem of how computers can automatically improve at certain tasks given data. The usefulness and range of applicability of such techniques has surged in the last decades due to the increase on accessible computational power and the amount of useful data available. In this section, a general overview of machine learning methods as well as the main tasks that can be addressed with them will be provided. Subsequently, the technical basis of two specific types of machine learning methods used in the next chapters will be explored: boosted decision trees and neural networks. Last but not least, we will go through a brief review of the common past use cases of these techniques at high energy physics experiments, especially focussing on those cases where they can be used to address some of the statistical inference and modelling issues from Chapter 3.