Resources Contact Us Home
Recurrent neural networks teaching system

Image Number 7 for United States Patent #5182794.

A teaching method for a recurrent neural network having hidden, output and input neurons calculates weighting errors over a limited number of propagations of the network. This process permits the use of conventional teaching sets, such as are used with feedforward networks, to be used with recurrent networks. The teaching outputs are substituted for the computed activations of the output neurons in the forward propagation and error correction stages. Back propagated error from the last propagation is assumed to be zero for the hidden neurons. A method of reducing drift of the network with respect to a modeled process is also described and a forced cycling method to eliminate the time lag between network input and output.

  Recently Added Patents
Electronic device
Ice tray for refrigerator
Temporary protective cover for an exposed junction box
Content reproduction apparatus and content processing method therefor
Light-reflective anisotropic conductive paste and light-emitting device
ZnO green luminescent material and its preparation
Phospholipid-based powders for drug delivery
  Randomly Featured Patents
Sialogogue, oral composition and food product containing the same
MILS network using COTS switches
Apparatus for sensing relative rotation between members
Two-piece interlocking corner protector
Hydrophilic synthetic gels and their use in electrophoresis
Multiple head floor sander
Sway control devices utilizable in conjunction with equalizer hitches
Drag reduction agent
Process for fabricating a lithographic mask