Main Page | Modules | Namespace List | Class Hierarchy | Class List | Directories | File List | Namespace Members | Class Members | File Members | Related Pages

Teem::BackPropFeedForwardNeuralNetwork Class Reference
[Controllers and neural networks]

#include <FeedForwardNeuralNetwork.h>

Inheritance diagram for Teem::BackPropFeedForwardNeuralNetwork:

Teem::FeedForwardNeuralNetwork Teem::Controller List of all members.

Detailed Description

Feed forward neural network with online back-propagation.

Use that way:

for i in inputs
    nn.setInput(i, inputValue);
nn.step();Ishtar::Variable<bool> outputLayerRecursive; 
for i in outputs
    nn.setError(i, errorValue);

Public Member Functions

 BackPropFeedForwardNeuralNetwork (size_t inputCount, size_t outputCount, const std::string &root)
 Constructor, create a feed forward neural network with online back-propagation with inputCount inputs and outputCount outputs.
virtual ~BackPropFeedForwardNeuralNetwork ()
virtual void stepBackward ()
 Backpropagate the error on the weight.
void setError (size_t index, double val)
 Set the desired output (used for back-propagation of error).
void setLearningRate (double rate)
 Set the learning rate value.
double getLearningRate ()
 Get the learning rate value.

Static Protected Member Functions

static double TanhBackwardActivationFunction (double x, double b)
 Tanh backward activation function.

Protected Attributes

Ishtar::Variable< double > learningRate
 learning rate constant
std::valarray< double > error
 desired output
ActivationFunction backwardActFunc
 derivative of the activation function
std::vector< std::valarray<
double > > 
 deltas (see backprop algorithm)

The documentation for this class was generated from the following files:
Generated on Mon Oct 24 17:38:27 2005 for Teem by  doxygen 1.4.2