Optical neural networks

We present an all-optical neuron by use of a multimode laser diode that is subjected to external optical feedback and light injection. The shape of the threshold function, that is needed for neural operation, is controlled by adjusting the external feedback level for two longitudinal cavity modes of the laser diode individually. One of the two modes corresponds to the output of the neuron, light injection at the wavelength of this mode provides excitatory input. Light injection in the other mode provides inhibitory input. When light corresponding to two input signals is injected in the same mode, summation of input signals can be achieved. A rate-equation model is used to explain the operating principle theoretically. The proposed injection seeding neuron is built using free-space optics to demonstrate the concept experimentally. The results are in good agreement with the predictions from the rate-equation model. Some experimental results show threshold functions that are preferable from a neural-network point of view. These results agree well with injection locking theory and experiments reported in literature.

show abstract

“…For mode we can write (2) where is the speed of light inside the laser material, is the internal cavity loss and the cavity length.…”

Section: A Rate-equation Analysismentioning

confidence: 99%

Optical neuron by use of a laser diode with injection seeding and external optical feedback

Mos

Hoppenbrouwers

Hill

et al. 2000

IEEE Trans. Neural Netw.

View full text Add to dashboard Cite

show abstract

“…where K5 = "ki (5) is a scalar product of the input vector X by the pattern çoS. Consequently, the output vector is a superposition of learning vectors with weights equal to their inner products input vectors.…”

Section: The Optical Neural Network Based On the Thin Holograms Applimentioning

confidence: 99%

“…The latter makes it possible to combine a large number of neurons with a complete set of interconnections. When neurons are assembled into nxn matrices (images) , the interconnections are described by a fourth-rank tensor and the dimension of a network with a full totality of interconnections has the form n2: n4: n2 [5] . In real life, however, a complete totality of links is not easy to implement [6] .…”

Section: Introductionmentioning

confidence: 99%

<title>Holographic correlator with thin holograms as an optical neural network</title>

Dubrovskaya

Shubnikov

1993

SPIE Proceedings

View full text Add to dashboard Cite

11 is shown that a holographic correlator with thin holograms application can be regarded as an optical neural network based on the outer product model. This network has a complete array of interconnections. Two versions of optical neural networks based on holographic correlators with interconnections, fixed in different places are presented. 1 . Introduction Numerous models of neural networks (NN) have been proposed as research and application tools. The most advanced model is the external product algorithm of McCulloch and Pitts [1] which has been extended by Metropolis et al. who introduced fluctuation dynamics and threshold switchings of neurons [2] . The algorithm has been further improved by Grossberg and Hopfield [3, 4]. A discrete model is described by the equation Y=NL(TX),where : and V are vectors which describe the neuron states of the input and output layers, T is the interconnection tensor which stores the learning results, and NL is a nonlinear operator. The energy of such a system is E=-2TXY, (2) and the interconnection array is described by a set of learning patterns (3) A continuous NN model which is not essentially different from a discrete model is described by nonlinear Lagrange differential equations and recognise the limited responses of neurons and interconnections and fluctuating noise sources. Technically, the structure of such a network can be represented as an input and an output layer of neurons, each connected with all the others. The network learns by using a set of patterns which stand for some interconnections with specified weights. When it is subjected to the action of an input stimulus, an NN relaxes towards 0819442244/93/$6.OO SPIE Vol. 1978 / 237 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/24/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

show abstract

“…18,19 A similar idea can be found behind the prototype optoelectronic neural-network implementations, based on optical matrix-vector multipliers, 20,21 which promise a large number of parallel processing elements. 22 The main disadvantage is that they suffer from some of the same problems as optical correlators (optical alignment, compactness and reconfigurability by means of SLM).…”

Section: Introductionmentioning

confidence: 99%

Optoelectronic neural processor for smart vision applications

Lamela¹,

Ruiz-Llata²

2007

The Imaging Science Journal

View full text Add to dashboard Cite

The implementation of a vision system in which the processing core is based on an optoelectronic neural network design is described. The vision system captures an image using a complimentary metal-oxide semiconductor (CMOS) image sensor, and the optoelectronic neural processor performs the classification among a set of sample patterns. The neural-network hardware architecture is based on an optical broadcast of inputs to an electronic array of processing elements, one per class. The benefit of this optoelectronic architecture, and thus the vision application, is that it is potentially scalable to a very large number of processing elements operating in parallel and at very high speed.

show abstract

Optical neural networks

Cited by 98 publications

References 35 publications

Optical neuron by use of a laser diode with injection seeding and external optical feedback

Optical neuron by use of a laser diode with injection seeding and external optical feedback

<title>Holographic correlator with thin holograms as an optical neural network</title>

Optoelectronic neural processor for smart vision applications

Contact Info

Product

Resources

About