Welcome to the Course

This is Advanced Machine Learning!

We’re very happy to welcome you to this course on Advanced Machine Learning. We hope you will enjoy it as much as we enjoyed developing it!

This course is a collaboration between the Persontyle Ltd, The Open University UK and the European Data Science Academy. The consortium has partnered with FutureLearn to provide you with this learning experience.

The Course Team: Who are we?

The lead educator for this course is Mike Ashcroft. He is chief AI officer at Persontyle, a lecturer at Uppsala University in Sweden and the founder of two companies that provide advanced machine learning services to industry, research groups and government.

The course facilitators will be Sophia Knight, who is a post-doc at Uppsala University, and Alex Yuan Gao and Lei You, both of whom are PhD students at Uppsala University.


We would like to know more about your motivation in joining our course. Please fill in our quick, optional pre-course survey.

Course Overview

In this course we will be looking at the following topics

Week 1

An overview of important background concepts and theory. This will provide the framework for you to locate the various algorithms and methods we look at in the remaining weeks.

Week 2

We look at how the evaluate the performance of statistical models, as well as the sufficiency of data (training, validation and test). We then look at the most popular supervised learning algorithms: Artificial neural networks.

Week 3

We continue with supervised learning algorithms, looking at the very important approaches known as kernel methods. After this, we look at unsupervised learning algorithms such as clustering and topic modeling. As part of the latter, we examine very important statistical learning algorithms such as Markov Chain Monte Carlo sampling algorithms and the Expectation Maximization algorithm.

Week 4

In the final week, we look at complicated aspects of pre-processing, such as feature engineering and dealing with missing data, before giving short introductions and examples of semi-supervised learning and reinforcement learning.


This course contains five different types of steps. Each sub-topic within the different area will typically consist of elements of the following

  1. An article, introducing the ideas and theory for the sub-topic. You should read and understand the contents of this article. There will occasionally be links to additional material which the interested student can follow, but such content is considered beyond the limits of this course.

  2. Videos providing additional information about the ideas and theory, or - more commonly - looking at coding exercises associated with the sub-topic. The latter videos give an opportunity for practical experience, and you can complete the exercises yourselves using R and the R script files accompanying these videos. The idea behind these code exercise videos is that you will see Dr Mike Ashcroft go through a script of code in R. He will describe what he is doing and how the code in working and you can replicate it if you like (using the associated code files). For those of you interested in getting practical practice, we really encourage you to work through these problems yourselves. If you are not interested in such practice, you can just listen to Mike explain what he has done and move on to the next step. To work with R in the same way as Mike does in the videos, you should download the R programming language and the R-Studio IDE. Both are free, as long as you choose the open source license for R-Studio. Note that time estimates for this course do not include time spent on these optional code exercises.

  3. Quizzes that allow you to check whether you understand selected important concepts from the article associated with the sub-topic.

  4. A discussion step where you can discuss the content from the articles, videos and quizzes covering a specific topic. Here you should ask questions if you don’t understand, or begin discussions on how these methods can be extended or applied, what their shortcomings and benefits might be or simple talk about the topics in any way that you are inclined to. Course educators will monitor these discussions to try and assist when there are common problems many students are encountering, and when other students appear unable to assist.

In addition, for those who wish to achieve a Certificate of Achievement, there will be weekly tests. They do not try to test knowledge of all ideas and algorithms covered, but rather test a randomly selected subset of important concepts. You will need to attempt every test question and score an average of 70% across all tests to achieve a Certificate of Achievement. For more information on Certificates of Achievement and how you can achieve one, please see our FAQs

Get extra benefits, upgrade your course

You can upgrade this course to obtain extra benefits, including:

Unlimited access to the course: Go at your own pace with unlimited access to the course for as long as it exists on FutureLearn.

Access to tests: Ensure you’ve mastered the material with access to tests on the course.

A Certificate of Achievement: To help you demonstrate your learning we’ll send you a Certificate of Achievement when you become eligible.

Find out more

Welcome and let’s get started!

Mike Ashcroft

Share this article:

This article is from the free online course:

Advanced Machine Learning

The Open University