How a network computes, why gradients vanish, and what makes depth trainable
One hidden layer solved XOR. It is fair to ask how far that goes. The answer is a theorem with a big name and a narrow content: a network with one hidden layer can come as close as you like to any continuous function on a closed interval, provided you are allowed to make the layer wide enough. This lesson builds the construction that proves it, then says carefully what the result does not promise.
🔒 This is a Pro lesson — the interactive figure, worked examples, quiz and practice open with Pro access.
▶ Universal Approximation