Backpropagation: The Step-by-Step Derivation
About this video
The backpropagation derivation, factor by factor with the chain rule.
Backpropagation applies the chain rule to share a network's error layer by layer. You first compute the error of the output layer, propagate it backward, and use it to obtain the gradients of every weight and bias in a single pass, summarised in four compact equations you can code directly.
Video transcript
- 0:01 The neuron combines input, weight and bias; the activation adds the
- 0:05 non-linearity and the loss measures the error.
- 0:08 The chain rule turns the derivative into three local factors:
- 0:12 the error, the slope of the activation and the input.
More videos
Explainer
The Computational Graph and Automatic Differentiation
The computational graph and automatic differentiation, in action.
Explainer
Backpropagation: The Intuition
The intuition of backpropagation: the error flows backward through the network.
Explainer
Batch, Stochastic and Mini-Batch Gradient Descent
Batch, stochastic and mini-batch: three ways to traverse the data.
