Now that we have got an insight into the possibilities to realize all-optical interconnection devices for different storage and interconnection purposes, we have to focus our attention on the second necessary key component of a neural structure: The nonlinear decision or thresholding element. The threshold unit is the key element of a neural net, because its slope decides, whether the net is able to solve nonlinear decision problems. Together with the interconnection unit, it forms the heart of any neural net and provides the means to realize simple optical neural networks, e.g. with associative memory features. Therefore, both components together are often called the optical neuron itself. If we are able to construct these two key compents of neural net devices, we are ready to construct already a variety of simple optical neural networks as they will be presented in the third part of that book.
KeywordsRing Resonator Threshold Function Spatial Light Modulator Phase Conjugate Soft Thresholding
Unable to display preview. Download preview PDF.
- 1.M. Wang, A. Freeman, Neural Function, Little, Brown, Boston (1987). This book describes in detail the different neuronal threshold functions, giving attention to problems as the realization of bipolar thresholding, exhibitory and inhibitory neuronal functions.Google Scholar