Floating Gaussian Mapping: a New Model of Adaptive Systems

Adaptive systems are usually realized by the neural network type of algorithms. In this contribu-tion an alternative approach, based on products of Gaussian factors centered at the data points, andacting as feature detectors, is proposed. Comparing with the feedforward neural networks with back-propagation learning is much faster because explicit construction of the approximation to the desiredmapping is performed, with fine tuning via subsequent adaptation of the shapes and positions of thefeature detectors. Comparing with the recurrent feedback networks this approach allows for full con-trol of the positions and sizes of the basins of attractors of the stationary points. Retrieving informa-tion is factorized into a series of one-dimensional searches. The FGM (Floating Gaussian Mapping)model is applicable to learning not only from examples but also from general laws. It may serve as amodel of associative memory or as a fuzzy expert system. Examples of application include identifica-tion of spectra and intelligent databases (associative memory type), analysis of simple electrical cir-cuits (expert system type), and classification problems (two-spirals problem).