From a single neuron to a network that learns
A story-driven, from-first-principles walk through neural networks — starting with a single artificial neuron and ending with a fully trained multi-layer perceptron, backpropagation, optimizers, and everything in between.
5 chapters · 64 min
Separating the hype from the actual idea
Teaching computers by example instead of by rule
What makes 'deep' learning different
Two winters, one thaw, and a revolution
How much of the brain analogy is actually true?
3 chapters · 40 min
4 chapters · 73 min
Vectors, matrices, and why networks are just matrix math
Derivatives, gradients, and the chain rule that makes learning possible
Making sense of uncertainty in data and predictions
How a network actually searches for good weights
8 chapters · 150 min
Layers, neurons, and how information flows forward
How a network learns from its own mistakes
The spark that lets networks learn anything non-linear
Teaching a network what 'wrong' means
Smarter ways to walk downhill
Why the starting point of training matters so much
Keeping a network from memorizing instead of learning
Putting every piece together into one training loop
4 chapters · 53 min
1 chapters · 8 min
Separating the hype from the actual idea