Learning internal representations by error propagation
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
A new approach to optical information processing based on neural network models with application to object recognition
| Hi-index | 0.00 |
Photonic implementations of neural networks, which use electronics to furnish gain and implement neural transfer functions and establish weighted connections between neutrons using incoherent light, are discussed. Fully or partially optical implementations incorporate coherent light and volume or planar holograms to establish interconnection weights, and spatial light modulators to implement neural transfer functions. The implementation of learning algorithms on optoelectronic neural networks is also discussed.