Machine Learning Techniques and Methods

Syllabus

• Week 1

  Introduction to machine learning and regression

  • Welcome to the course

    Welcome to Machine Learning Techniques and Methods. Let's start by finding out what you’ll be learning this week, as well as looking at some heart data to get our minds in gear.

  • What are the principles of regression?

    In regression, we consider the problem of predicting a label that can take on continuous values. Here, we'll define regression with basic mathematical notation. Then, we'll discuss the role of optimisation in regression problems.

  • Basic regression models

    Next, we will introduce the mean squared error, which we use to quantify the quality of a regression model. Then, we will discuss linear models for simple and multiple regression. Finally, we will look at polynomial models.

  • Exploring model validation

    Next, we will look into the ability of our trained models to generalise. We'll introduce 'underfitting' and 'overfitting' to describe this ability, and connect them to the notion of complexity. We'll also discuss model validation.

  • Weekly wrap-up

    This will conclude the first week of the course and we will take a sneak peek of what we will be seeing in Week 2.

• Week 2

  Classification

  • Welcome to Week 2

    Last week, we discussed the first family of supervised problems: regression. This week we will delve into classification, the second of these families, that can be found in many scientific, engineering and business scenarios.

  • What are the principles of classification?

    In classification we consider the problem of predicting a label that can take on discrete values. In these steps, we will use basic mathematical notation to define classifiers and look at their main components.

  • Basic classification models

    In these steps, we will explore two popular families of classification models: linear classifiers and nearest neighbours. We will also discuss the notions of overfitting and complexity in the context of classification.

  • Exploring classification performance

    Next, we will look at the importance of evaluating classifiers. We'll then introduce the concept of the confusion matrix and explore it further in a Python notebook.

  • Weekly wrap-up

    In these steps, we will recap what we have learnt during this week and introduce the topic of next week.

• Week 3

  Clustering

  • Welcome to Week 3

    Welcome to Week 3 of Machine Learning Techniques and Methods. Let’s start by discussing a problem involving data. Then, you’ll find more about what you’ll be learning this week.

  • What are the principles of clustering?

    In clustering, we consider the problem of identifying groups of samples within our dataset. In these steps, we will define clustering using a notion of similarity between samples based on their location in the attribute space.

  • Basic clustering models

    Next, we'll introduce the intra-cluster and inter-cluster sample scatter, which we can use to quantify the quality of a clustering solution. Then, we'll discuss K-means clustering. Finally, we'll look at density-based clustering.

  • Exploring clustering further

    Next, we will cover some practical considerations about clustering, such as specifying the number of clusters in K-means. We will also discuss hierarchical clustering approaches.

  • Weekly wrap-up

    This will conclude the third week of the course, and will introduce the fourth and last week of the course, focused on revision of contents and preparing to work on the final coursework.

• Week 4

  Revision and assessment preparation

  • Welcome to Week 4

    Welcome to the fourth and final week of the second course. We will revise the contents learnt throughout the previous three weeks and we will describe the assessment for this course.