Optical neural network apparatus using primary processing

Claim 1

An optical neural network apparatus using a primary processing, comprising: (a) a primary processing device for subjecting two-dimensional image input information to primary processing to extract the characteristic feature of an object pattern; and (b) an optical neural network device for parallely processing compressed information adaptively with optical computing; wherein said optical neural network device is an optical associative memory device which includes: a multiply imaging system for multiply imaging a characteristic feature extraction image read from said primary processing device; an input pattern conversion device for holding an image thus multiply imaged; a memory matrix holding device for holding a memory matrix in response to a cross correlation between an input pattern and an associative output pattern; a locally imaging system for locally imaging an image formed by sequentially reading images held by said input pattern conversions device and said memory matrix holding device to implement partial sum computation through optical computing; an output function computation device for applying an output function to an image thus locally imaged through the optical computing and obtaining an associative output; a magnifying/imaging system for magnifying and imaging a learning output pattern or an associative output pattern for an input pattern to be stored; and a learning/computation device for learning said memory matrix through optical computing with use of an image thus magnified and imaged and the image multiply imaged by said input pattern conversion device.