This is BackPropEngine.rtf in view mode; [Download] [Up]
BPE V1.0 29-SEP-1991 Ralph Zazula zazula@pri.com BackPropEngine INHERITS FROM Object DECLARED IN BackPropEngine.h CLASS DESCRIPTION This class implements a feed-forward network that updates it's weights using the Error Back-propagation algorithm. INSTANCE VARIABLES Inherited from Object Class isa; Declared in BackPropEngine id inputs; id hidden; id outputs; float ETA; float ALPHA; inputs a List of the input Neurons hidden a List of the hidden layer Neurons outputs a List of the output Neurons ETA the learning-rate ALPHA the momentum-term (not used) METHOD TYPES Initializing a new instance ± init ± initWithInputs:hidden:outputs: Running the algorithm ± applyInput: ± correctWithTarget: Getting parameters ± inputs ± hidden ± outputs INSTANCE METHODS applyInput: - applyInput:(double *)input Feeds the input vector pointed to by input forward through the BP network. correctWithTarget: - correctWithTarget:(double *)target Modifies the weights of the BP network by comparing the current output to the target vector target. This assumes that you have used the applyInput: method to feed-forward an input vector. See also: - applyInput: hidden - hidden Returns a pointer to the receivers list of hidden Neurons. init - init Initializes the receiver. This should not be called directly. Use initWithInputs:hidden:outputs: to initialize a new BackPropEngine. See also: ± initWithInputs:hidden:outputs: initWithInputs:hidden:outputs: - initWithInputs:(int)Nin hidden:(int)Nhid outputs:(int)Nout Initializes the receiver creating a feed-forward BP network with Nin inputs, Nhid hidden nodes and Nout outputs. See also: - init inputs - inputs Returns a pointer to the receivers list of input Neurons. outputs - outputs Returns a pointer to the receivers list of output Neurons.
These are the contents of the former NiCE NeXT User Group NeXTSTEP/OpenStep software archive, currently hosted by Netfuture.ch.