Adaptive Optical Neural Network Classifier Systems

摘要: Abstract : We present two adaptive opto-electronic neural network hardware architectures capable of exploiting parallel optics to realize real-time processing and classification high-dimensional data. The are based on radial basis function networks that employ on-line training techniques offer robustness noise optical system imperfections. A binary-input data classifier is presented first the issues imperfections, device characterization, addressed. experimental results from compared with a computer model in order identify critical sources indicate possible areas for performance improvements. grayscale-input then proposed handling 8 bit input broaden range applications classifier. An wavelet transform intended use as multi-resolution image preprocessor classifiers discussed.