Neural networks and gradient descent

Neurons, layers, loss functions and backpropagation, built with toy examples.

5 lessons planned · none written yet

  1. 1A neuron is a weighted voteNot written yetWeights, bias and nonlinearity.
  2. 2LayersNot written yetWhy depth helps.
  3. 3Loss and gradient descentNot written yetLearning as walking downhill.
  4. 4BackpropagationNot written yetThe chain rule, applied a billion times.
  5. 5Training in practiceNot written yetBatches, learning rates, overfitting.