Deep Learning

Neural networks from first principles. Build the intuition first, one chapter at a time.

What Is a Neuron?
Weights, bias and activation functions, built from scratch, with four figures you drive yourself. Ends at the point where one neuron stops being enough.
How Machines Learn by Bending Space
A network doesn’t learn a complicated boundary — it folds the space until a straight one will do. Eight figures, ending with loss, gradient descent and backpropagation.