next up previous contents index
Next: Parameters of SCG Up: Scaled Conjugate Gradient Previous: Conjugate Gradient Methods

Main features of SCG

Let be a vector from the space , where N is the sum of the number of weights and of the number of biases of the network. Let E be the error function we want to minimize.

SCG differs from other CGMs in two points:

SCG has been shown to be considerably faster than standard backpropagation and than other CGMs [Mol93].
Tue Nov 28 10:30:44 MET 1995