---

TestNN Class Reference

Inheritance diagram for TestNN:

List of all members.

Public Methods
	TestNN ()
virtual void	startData ()
	Called by train() to signal the start of a cycle thru the data. This function should reset any indices in preparation for calls to nextDatum().
virtual bool	nextDatum ()
	Called by train() to request that the next training point be set up on the inputs and outputs of the NN. Return true if a new datum was set up. Return false if a datum is not available, indicating the end of the data set. Similar semantics to end of file flag.
virtual bool	correct ()
	Determines if the outputs are correct (by whatever contorted method). Called by train() to support the construction of "happy graphs", but not necessary to implement.
virtual void	happyGraph (int iteration, float accuracy)
	Called by train() to support user level construction of learning curve graph. "iteration" counts the number of cycles thru the data so far, and "accuracy" is a number between 0 and 1, where 1 is perfect.
Public Attributes
int	index
int	dataCount
char *	data [9000]
char	buffer [80]

---

Constructor & Destructor Documentation

TestNN::TestNN (    )  [inline]

---

Member Function Documentation

virtual bool TestNN::correct (    )  [inline, virtual]

Determines if the outputs are correct (by whatever contorted method). Called by train() to support the construction of "happy graphs", but not necessary to implement. Reimplemented from fl::NeuralNetworkBackprop.

virtual void TestNN::happyGraph (  int    iteration, float    accuracy )  [inline, virtual]

Called by train() to support user level construction of learning curve graph. "iteration" counts the number of cycles thru the data so far, and "accuracy" is a number between 0 and 1, where 1 is perfect. Reimplemented from fl::NeuralNetworkBackprop.

virtual bool TestNN::nextDatum (    )  [inline, virtual]

Called by train() to request that the next training point be set up on the inputs and outputs of the NN. Return true if a new datum was set up. Return false if a datum is not available, indicating the end of the data set. Similar semantics to end of file flag. Implements fl::NeuralNetworkBackprop.

virtual void TestNN::startData (    )  [inline, virtual]

Called by train() to signal the start of a cycle thru the data. This function should reset any indices in preparation for calls to nextDatum(). Implements fl::NeuralNetworkBackprop.

---

Member Data Documentation

char TestNN::buffer[80]

char* TestNN::data[9000]

int TestNN::dataCount

int TestNN::index

---

The documentation for this class was generated from the following file:

• testNN.cc

---