Optical correlator production system neural net

Algorithms for optoelectronic implementation of modified signed-digit division, square-root, logarithmic, and exponential functions

Algorithms for computing complex elementary functions based on the modified signed-digit (MSD) number system representations are proposed. An arithmetic unit that performs parallel one-step addition (subtraction), multiplication, and division is proposed to achieve the computation of complex functions such as the square-root, logarithm, exponential, and other related operations. An optoelectronic correlator-based architecture is suggested for implementing the proposed MSD algorithms to compute the above-mentioned elementary functions. We utilized the symbolic substitution technique to reduce the number of computation rules involved.

Designs of optoelectronic trinary signed-digit multiplication by use of joint spatial encodings and optical correlation

Trinary signed-digit (TSD) symbolic-substitution-based (SS-based) optical adders, which were recently proposed, are used as the basic modules for designing highly parallel optical multiplications by use of cascaded optical correlators. The proposed multiplications perform carry-free generation of the multiplication partial products of two words in constant time. Also, three different multiplication designs are presented, and new joint spatial encodings for the TSD numbers are introduced. The proposed joint spatial encodings allow one to reduce the SS computation rules involved in optical multiplication. In addition, the proposed joint spatial encodings increase the space-bandwidth product of the spatial light modulators of the optical system. This increase is achieved by reduction of the numbers of pixels in the joint spatial encodings for the input TSD operands as well as reduction of the number of pixels used in the proposed matched spatial filters for the optical multipliers.

Optoelectronic Recoded and Nonrecoded Trinary Signed-Digit Adder that uses Optical Correlation

A symbolic-substitution-based optical numeric processor that uses recoded and nonrecoded trinary signed-digit (TSD) number representations is proposed. Also, we propose new joint spatial encodings for the TSD numbers that reduce the symbolic-substitution computation rules involved in the processor. Optoelectronic implementation of the proposed recoded adder is feasible. Also, the nonrecoded TSD addition can be performed optically in two steps. Both the proposed recoded and nonrecoded adders are more compact than a recently reported modified signed-digit counterpart and use fewer correlators and spatial light modulators.

Neural network implementation using self-lensing media

An all-optical implementation of a feed-forward artificial neural network is presented that uses self-lensing materials in which the index of refraction is irradiance dependent. Many of these types of material have ultrafast response times and permit both weighted connections and nonlinear neuron processing to be implemented with only thin material layers separated by free space. Both neuron processing and weighted interconnections emerge directly from the physical optics of the device. One creates virtual neurons and their connections simply by applying patterns of irradiance to thin layers of the nonlinear media. This is a result of a variation of the refractive-index profile of the self-lensing nonlinear media in response to the applied irradiance. An optical-backpropagation training method for this network is presented. The optical backpropagation is a training method that can be implemented potentially within the same optical device as the forward calculations, although several issues crucial to this po sibility remain to be addressed. Such a network was numerically simulated and trained to solve many benchmark classification problems, and some of these results are presented. To demonstrate the feasibility of building such a network, we also describe experimental work in the construction of an optical network trained to perform a logic XNOR function. This network, as a proof of concept, uses a relatively slow thermal nonlinear material with ~1-s response time.

Optoelectronic implementation of a fuzzy winner-take-all network for fuzzy logic inference

A fuzzy winner-take-all model is proposed for performing fuzzy logic inference. An optoelectronic scheme for this model is presented, and experimental results are given.

Parallel electro-optical rule-based system for fast execution of expert systems

The slow execution speed of current rule-based systems has restricted their application areas. Multiprocessor architectures have been proposed to overcome this limitation. However, as the number of processors in a multiprocessor system grows, so does the cost of communication between processors or between processor and memory units. The use of optics for a fast and parallel implementation of rule-based systems is proposed. The proposed optical system is hybrid in nature, using electronics for the user interface and optics for the rule-based inference engine. The proposed system uses twodimensional planes as basic computational entities and is therefore able to provide concurrent rule processing. Furthermore, it provides highly efficient implementation of the basic operations needed in rule-based systems; namely, matching, selection, and rule firing. The execution speed of the proposed system is theoretically estimated and is shown to be potentially of orders of magnitude faster than current electronic system.