What we will cover in the course

Week by week

The main topics we will cover during this five-week course are as follows:

Week 1 – introduction to deep learning:

• Recap of core machine learning concepts
• Machine learning versus deep learning
• Deep learning libraries – PyTorch

Week 2 – convolutional neural networks (CNNs):

• Network components – convolutional layers, max pooling layers, activation functions
• Tensors and dimensionality
• Bulding a simple CNN

Week 3 – training CNNs:

• The training loop
• Hyperparameters – improving the training algorithm
• Datasets and dataloaders – managing your data

Week 4 – classification and regression:

• Encoder architectures
• Making an image classifer
• Regression using deep learning

Week 5 – Spatial approaches and more:

• Encoder-decoder architectures
• Image segmentation
• Heatmap regression
• Bounding box methods
• Multi-task learning

In addition to the above topics there will also be a series of practical exercises where you can get some hands on experience of developing simple deep learning models. Don’t worry – all the code can be run in a browser window and we will give all the instructions you need as we go.

Week 1 in more detail

In week one will set the scene for the course as a whole by talking about what deep learning is, what problems it can try and solve, and how it differs from more traditional machine learning approaches you may have encountered previously.

We will also give a brief introduction to the important software tools we will use during the course, namely Python PyTorch and Gogle Colab, as well as a sort discussion about the use of GPU acceleration – of vital importance for deep learning projects of any size.

Week 1 learning outcomes

• Compare machine learning with deep learning
• Practice the use of GPU acceleration with Google Colab