Course: Machine Learning for Image Data

What is a neural network model? This video goes through the basics of nodes, input layers, output layers and hidden layers, and how nodes in the network are activated.

How does deep learning differ from machine learning? Broadly speaking, in traditional machine learning you will need to select both a type of model and choose which features to extract …

This week we will look at deep learning. In particular we will consider where it comes from, and how it builds on existing machine concepts and techniques we’ve looked at …

This week, among other things, we have looked at expanding dataset size using data augmentation, and the use of cross-validation during model evaluation. In this week’s practical, we will demonstrate …

Well done for completing Week 4 of the course. This week we looked at: things to think about when training models, including: dataset size – use of performance curves splitting …

What do we mean by class imbalance? If your dataset has lots of examples of one class, but very few of another, then there is said to be class imbalance. …

In general, the more features or dimensions our dataset has, the more sparsely they are positioned in that dimensional space. This video explains the so-called curse of dimensionality in more …

How can we augment our datasets? In this video we consider a few ways including: flipping adding noise. We also discuss the pros and cons of different data augmentation approaches.

Cross validation is a common method used to evaluate machine learning models on all parts of the data set. This video explains how it works in more detail.

What is data augmentation? Data collection can be either time or resource expensive (or both!), and so often machine learning datasets can be augmented by adding artificial variations of individual …

How do we go about splitting our datasets? Previously we’ve seen how it’s important to split datasets into different subsets for training and validation. This video discusses how we might …

Good performance of machine learning models depends on good training practice. In this video we consider the use of use of training and validation datasets and how the performance of …

Welcome to Week 4 of the course. This week we will look at some tips and tricks you may need to consider when using machine learning on your data. In …

This week we will look in more detail at methods for: clustering classification regression. Plus, we will also look at ways to perform: model evaluation model visualisation model selection.

In this video, we take a closer look at overfitting and how to avoid it. We will look at: what is overfitting and what causes it? model complexity bias and …