|
|
Radial networks are usefull in classificational methods,
detecting damages in some systems, recognition of
pattern. Application of radial networks in predictions
of complicated time series and prediction of monthly
changed emloyment level, economical trends lets to
obtain very good results. |
|
Radial networks are
natural completion of sigmoidal networks. Sigmoidal
neuron represents in multidimensional space hiperplane,
separating that space to two categories, however radial
neuron represents hipersphere, by radial separation
around central point. |
|
|
Back - Main
|
|
|
|
|
| |
|