Optoelectronic Neural Networks

Conclusions

Results of experiments show that neuron used for primary processing limits number of output data from the sensor system but such processing limits also applications of the method to the cases in which only threshold is important. The problem may be solved by use more advanced neural processing in which multithreshold system for each sensor could be applied.

System of neuron operation requires different levels of optical output power. The level values of sensing variables differ in geometrical progression. Hence number of sensors working with optoelectronic neuron as a processor are limited and depends of transmission characteristics of LC cell or each pixel of spatial light modulator. The problem may be particularly solved when the cascade of neuron layers creating a neural network are used.