Deep Learning
How can machines automatically learn representations?
Build from a single artificial neuron to networks that learn useful internal features.
1. Perceptron
The simplest possible artificial neuron
2. Neural Networks
Combine many perceptrons into hidden layers
3. Activation Functions
Add non-linearity so stacked layers can learn richer functions
4. Backpropagation
Compute gradients through a network with the chain rule
5. Training Deep Networks
Train neural networks effectively with modern optimisation techniques
6. Convolutional Neural Networks
Learn spatial patterns in images with filters and feature maps
7. Sequence Models
Process ordered data with recurrent neural networks
8. LSTMs and GRUs
Remember information across longer sequences
9. Attention
Let tokens choose which other tokens matter
10. Transformers
Use attention as the primary computation