Next: Step Function Previous: Example

Update Function

Why use the derivative of the transfer function (the sigmoid) in the calculation of delta?

Note the visualization of the weight space using gradient descent.

We want to move the weights in the direction of steepest descent in this space.

To do this, compute derivative of the error with respect to each weight in the equation. This results in the delta terms showed earlier.

Note that the transfer function must be differentiable everywhere.