· advanced
Deep Learning Foundations
What a neural network is actually doing, why gradient descent works, and when a network is the right answer rather than the fashionable one. Enough to read a training script and know why it is not converging.
Written in house, from the work rather than from a syllabus. Every example is one we would use ourselves, and every quiz question explains why its answer is right.
What you’ll learn
- Describe what a layer computes and why a non-linearity is required
- Explain gradient descent and what the learning rate controls
- Read a loss curve and name the likely fault
- Choose batch size and epochs with an understanding of the trade
- Apply regularisation, and recognise overfitting in a network
- Use a pretrained model, and say when training from scratch is wrong
Before you start
- · Machine Learning Basics and Model Evaluation, or equivalent
- · Comfort with the idea of a function and a derivative