Teaching: 15 min
Exercises: 0 min
  • What is machine learning?

  • What role does machine learning have in high energy physics?

  • What should I do if I want to get good at machine learning?

  • Discuss the general learning task in machine learning.

  • Provide examples of machine learning in high energy physics.

  • Give resources to people who want to become proficient in machine learning.

What is Machine Learning?

General definition

In a traditional approach to solving problems, one would study a problem, write rules (i.e. laws of physics) to solve that problem, analyze errors, then modify the rules. A machine learning approach automates this process: a machine learning model modifies its rules based on the errors it measures. This abstract statement becomes more conceptual once the mathematical foundations are established. For now, you can think of the following example:

There are three main tasks in machine learning:

  1. Regression. The input is multi-dimensional data points and the output is a real number (or sequence of real numbers). For example, the input might be height and weight, whilst output might be resting heart rate and resting blood pressure (a sequence of real numbers).

  2. Classification. The input is multi-dimensional data points and the output is an integer (which represents different classes). Consider the following example with two classes: pictures of roses and pictures of begonias. The input would be multi-dimensional images (color channel included) and one may assign the integer 0 to roses and the integer 1 to begonias.

  3. Generation: The input is noise and the output is something sensible. For example, training a machine learning algorithm to take in a random seed and generate images of peoples faces. Another example could be ageing filters on social media.

What Role Does Machine Learning have in High Energy Physics?

Machine learning has become quite popular in scientific fields in the past few years. The following plot from MIT Technical Review examines the number of papers available in the “artificial intelligence” section of arXiv up to November 18, 2018.

ML Popularity

The following are a few recent articles on machine learning in high energy physics

Where to Become Proficient in Machine Learning

Machine Learning is not something you’ll learn in an hour. It’s a skill you need to develop over time, and like any skill you need to practice a little bit every day. If you want to really excel at machine learning, my recommendation is to try an online course such as this one. If textbooks are your thing, try this book, ensuring to read and code along with each chapter. Don’t go crazy: just do 30 minutes a day. You’d be surprised how much you could learn in a couple months. In summary:

Just for a bit of perspective, I started learning about machine learning in April 2019. Don’t expect the learning process to be a quick one: follow online courses and code along with them. If you have a textbook, read through it thoroughly and make sure you code along with the textbook.

Your feedback is very welcome! Most helpful for us is if you “Improve this page on GitHub”. If you prefer anonymous feedback, please fill this form.

Key Points

  • The 3 main tasks of Machine Learning are regression, classification and generation.

  • Machine learning has many applications in high energy physics.

  • If you want to become proficient in machine learning, you need to practice.