public interface MLAutoAssocationextends MLRegressionDefines a MLMethod that can handle autoassocation. Autoassociation is a simple form of pattern recognition where the MLMethod echos back the exact pattern that the input most closely matches. For example, if the autoassociative MLMethod were trained to recognize an 8x8 grid of characters, the return value would be the entire 8x8 grid of the character recognized. This is the type of recognition performed by Hopfield Networks. It is also an optional recognition form used by GR/PNN's. This is a form of unsupervised training.
SCaVis 1.7 © jWork.org