Backpropagation of errors

• Use a nonlinear, differentiable activation function

  • Such as a sigmoid

• Use a multilayer feedforward network

  • Outputs are differentiable functions of the inputs

• Result: Can propagate blame back to internal nodes

  • Chain rule (calculus) gives Dwij for internal nodes
  • Use a gradient-descent learning rule

Previous slide

Next slide

Back to first slide

View graphic version