PatentDe  


Dokumentenidentifikation EP0717366 19.09.2002
EP-Veröffentlichungsnummer 0717366
Titel Neuronales Netzwerk mit einbegriffenen direkten optischen Bildern
Anmelder Xerox Corp., Rochester, N.Y., US
Erfinder Stearns, Richard G., Los Gatos, US
Vertreter derzeit kein Vertreter bestellt
DE-Aktenzeichen 69527767
Vertragsstaaten DE, FR, GB
Sprache des Dokument EN
EP-Anmeldetag 12.12.1995
EP-Aktenzeichen 953090537
EP-Offenlegungsdatum 19.06.1996
EP date of grant 14.08.2002
Veröffentlichungstag im Patentblatt 19.09.2002
IPC-Hauptklasse G06K 9/42
IPC-Nebenklasse G06E 3/00   G06K 9/82   

Beschreibung[en]

This application pertains to the art of neural networks, and more particularly to an optically addressed neural network according to the preamble of claim 1.

The invention is particularly applicable to extraction of information from visual images and will be described with particular reference thereto. However, it will be appreciated that the invention has broader applications, such as in efficient fabrication of any neural network.

A concept similar to the one as outlined in the preamble of claim 1 is taught in the article Optically programmed neural network capable of stand-alone operation, Applied Optics, 32(19993) September 10, No 26, by Richard G. Stearns.

The field of artificial intelligence-based computing is expanding rapidly. Many of the fundamental concepts that have been developed in the last few years, are already reaching the stage of second or third generation in sophistication. One such area is the subdivision of artificial intelligence comprising neural networks.

A first generation neural network requires functional, digital computing units which were interconnected by weighting values. Such functional units could be comprised of independent hardware processors, or alternatively implemented by software. In either instance, complex and extensive digital computing power was required.

More recently, it has been recognized that neural networks may be realized by a combination of electronics and optics. In such systems, a combination of photoconductive elements and lighting applied thereto was implemented to create a neural network. See, for example, Stearns, Richard G., Trainable Optically Programmed Neural Network, Applied Optics, Vol. 31, No. 29, October 10, 1992, pp. 6230-6239.

In this system, operation of a neural network with optically-addressed weighting, constructed from a two-dimensional photoconductor array that is masked by a liquid-crystal display (LCD) was provided. A fully trainable three-layer perceptron network was demonstrated using this architecture, which was capable of operating in a completely standalone mode, once trained. In the previous work, data was input to the network electronically, by applying voltages to the photoconductor array.

Thus, it is the object of the present invention to provide a network architecture that may be extended to allow direct optical input.

According to the present invention this object is solved by any of the optically addressed neural network having the features set out in claims 1 and 9, and by a neural method according to claim 10.

Once trained, the network is capable of processing in real time an image projected onto it, in a completely standalone mode.

Artificial neural networks appear to be naturally suited to a number of image processing problems, including for example pattern recognition. This results in part from their inherent parallel architecture, as well as their ability to perform well in the presence of image noise and degradation. It would follow that a compact hardware architecture that could combine optical image capture and neural network processing would be of significant interest. The subject application teaches such an architecture, which combines real-time image capture and neural network classification within a single processing module.

Preferably, the first subset is comprised of first and second rectangular sub-portions sharing no common column conductor therebetween. The neural network preferably further comprises a second rectangular subset of array elements sharing at least one of no column conductor and no common column conductor with one of the first and second rectangular sub-portions.

In accordance with a more limited aspect of the subject invention, the input lines are provided by an image capture mechanism disposed within the two-dimensional array of photoconductors itself.

Preferably, the mask includes a first window portion defining light passing therethrough so as to affect a subset wkj of the photoconductors, which subset is defined by a subset of the columns of the photoconductors in the J rows.

Preferably, the mask includes a second window portion defining light passing therethrough so as to affect a subset wlk of the photoconductors, which subset is defined by a subset of the columns of the photoconductors not disposed in the J rows. Preferably, the mask includes a generally opaque portion so as to prevent light from substantially affecting the photoconductors on all areas of the twodimensional array other than wkj and wlk. Preferably, the mask includes a third window portion so as to communicate an image to an area wji of the photoconductors, which area shares no column conductor or row conductor with the conductors wkj or wlk, and wherein column conductors associated with photoconductors of the area wji comprise the J input conductors. Preferably, photoconductors of the areas wkj and wlk share no column conductor and now row conductor. Preferably, the second subset is formed from amplified column signals associated with one-half the number of column conductors of wlk.

Regarding the neural method according to claim 10, preferably, the step of selectively controlling includes the step of selectively maintaining subportions of the mask at a substantially opaque transmissivity.

The method preferably further comprises the steps of: receiving an optical image into an imaging array of photoconductors; and communicating an output signal from the imaging array generated as a result of a received optical image to each of the array nodes of the first defined subset as the input signal. The method preferably further comprises the step of outputting a portion of the amplified signals, unique to the array feedback portion of the array nodes being unique to the first defined subset, as an output signal. The method preferably further comprises the step of impressing includes the step of impressing the amplified signals to the array feedback portion defined as unique to the first defined subset.

An advantage of the present invention is the provision of a highly efficient, compact neural network system.

Yet another advantage of the present invention is the provision of a highly-integrated image recognition system.

Yet a further advantage of the present invention is the provision of an accurate neural network which may be fabricated inexpensively.

The invention may take physical form in certain parts and arrangements of parts, a preferred embodiment and method of which will be described in detail in this specification and illustrated in the accompanying drawings which form a part hereof, and wherein:

  • FIGS. 1(a) and 1(b) illustrate, in schematic form, an architecture of a three-layer perception network of the subject invention;
  • FIGS. 2(a)-2(k) illustrate a modified forms of the architecture of FIG. 1 wherein an incident image is projected onto an array of photoconductors to accomplish specified neural functions;
  • FIGS. 3(a)-3(j) illustrate a series of examples in connection with a training set for handprinted digit recognition;
  • FIG. 4 illustrates a network training error for the example handprinted digit recognition;
  • FIGS. 5(a)-5(j) illustrate, in histogram form, a typical real-time classification performance of a trained network of the subject invention;
  • FIGS. 6(a)-6(h) illustrate several examples of images used to train the network to recognize a series of human faces;
  • FIG. 7 illustrates a suitable liquid crystal display pattern used to filter an optical image in training a network to recognize faces, such as was provided in FIG. 6;
  • FIG. 8 illustrates a network training error for facial recognition;
  • FIGS. 9(a)-9(j) illustrate examples of realtime classification of performance of a trained network; and
  • FIG. 10 illustrates a schematic of an alternative embodiment of a suitable system implementation of the subject invention.

FIG. 1 illustrates an optically-addressed neural network A of the present invention.

The key elements of the optically-addressed neural network have been described in detail elsewhere, and hence will only be briefly discussed here. The network is based upon the combination of a two-dimensional array of photoconductors, with an LCD that is aligned above the photoconductor array. The photoconductor array is then illuminated through the LCD. A schematic 10 of the two-dimensional photoconductor array 12 is shown in Fig. 1(a). It consists of a grid of 128 horizontal lines 14 and 128 vertical lines 16, with an interdigitated a-Si:H photoconductive sensor fabricated at each node 18 of the grid. Thus at each node 18 is a resistive interconnection, which is optically programmed by varying the light incident upon it through a generally aligned LCD panel 20. The system is suitably configured so that voltages are applied along the horizontal lines 14 of the array, with currents measured through the vertical lines 18, which are advantageously held at virtual ground potential. The vertical lines 18 are suitably paired to allow bipolar weighting in the neural network. The 64 pairs of vertical lines 16 are fed into 64 nonlinear differential transresistance amplifiers 30. The outputs of these 64 amplifiers are routed back to the bottom 64 horizontal conducting lines 24 of the sensor array (see Fig. 1b).

A pitch of the photoconductor array in both the horizontal and vertical directions is suitably 272 µm. An active-matrix LCD 20, whose pixel pitch is 68µm, is aligned in the preferred embodiment directly above the sensor array 12, so that groups of 4x4 LCD pixels mask each photoconductive sensor beneath. The LCD 20 and sensor array 12 are then illuminated by a relatively collimated light source, such as a common 35mm projector.

In Fig. 1(b) is shown schematically how the system is driven to implement a standard perceptron network with one hidden layer. The LCD array 20 is maintained in a minimum transmission state, except for two rectangular areas which correspond to the weight fields wkjI, between the input-layers 24 and hidden-layers 26, and wlkII, between the hidden-and output-layers of the network, respectively. In the system of Fig. 1(b) the network has J input neurons, which correspond to J voltages applied to a subset of the top 64 horizontal lines of the sensor array. There are K hidden-layer neurons, which because of the vertical line pairing, make use of 2K vertical lines on the sensor array. Thus the differential current that flows through each of the K vertical line pairs of the sensor array 12 correspond to the input of a hidden-layer neuron. The nonlinear transresistance amplifiers perform sigmoidal transformation of the input signals, and their output voltages correspond to the hidden-layer neurons output values. By varying the transmission of pixels of the LCD array 20 in the rectangular region wkjI, interconnections weights between the input-layer and hidden-layer neurons are adjusted.

A similar rectangular region (wlk) is defined between the K horizontal lines of the photoconductor array 12 which carry the voltages of the hidden-layer neuron outputs, and L vertical line pairs (2L vertical lines), which carry differential currents corresponding to the output-layer neurons. This second rectangular region wlkII defines the interconnections weights between the hidden-layer and output-layer neurons of the network. Output lines 24 are shown connected to a nonlinear current-to-voltage conversion unit or amplifier 30.

Because there are 16 LCD pixels above each photoconductive sensor of the LCD array 20, and two neighboring photoconductive sensors are used to produce one bipolar weight in the network, there are 32 LCD pixels used for each interconnection weight. The LCD is capable of generating -16 levels of gray per pixel, which should in theory allow a large number of weight levels. In the present embodiment, 129 levels of bipolar weighting are implemented, using a halftoning scheme that allows one LCD pixel of the 32 to be driven in a gray-scale mode at any given time, with all others driven in binary fashion. This approach is suitably taken to minimize the effect of LCD pixel nonuniformity, which is most pronounced when pixels are driven at intermediate gray values.

The LCD array 12 is suitably driven by a Sun IPX workstation. In addition, the top 64 lines of the sensor array are driven by digital-to-analog converters, which are suitably controlled by the same workstation. The workstation furthermore is interfaced to analog-to-digital conversion circuitry, which monitors the outputs of the 64 nonlinear transresistance amplifiers shown at the bottom of the sensor array 12 in Fig. 1(b).

A suitable range of voltages applied to the top 64 horizontal lines of the sensor array is ± 10V. The nonlinear transresistance amplifiers have a small signal gain of 2 x 108 V/A (differential input current), and saturate in a sigmoidal fashion at output levels of ± 10V. Associating the hardware network with a theoretical network whose neurons obey a tanh(x) transformation relation, it follows that a voltage of 10V in the hardware network corresponds to a neuron output value of 1.0 in the theoretical network. In typical operation, a single bipolar weight is found to produce a maximum differential current of 0.80 µA, resulting in a voltage of 7.4V at the transresistance amplifier output. This implies that a single weight in the hardware network is limited to a maximum value corresponding to tanh-1(7.4) = 0.95 in the corresponding theoretical network.

Training occurs by adjustment of the interconnection weights, using a standard backpropagation algorithm. As mentioned before, neuron output values of the hardware network are mapped to those of the standard theoretical network by equating ± 10V in the hardware network to outputs of ± 1.0 in the theoretical network. The interconnection weights of the hardware network are conveniently mapped to values between ± 1.0 in the theoretical network (within measurement error of the maximum measured values ± 0.95). To train the network, backpropagation of the hardware network output error is performed by the digital computer. In performing the backpropagation, the output values of the hidden-layer and output-layer neurons are measured directly, and an ideal tanh(x) neuron transfer function is assumed. The weight values used in the backpropagation are the intended values which have been programmed onto the LCD array 20. Thus any nonuniformity or nonideality in implementation of the weighting through the LCD, or in the nonlinear transformation of the transresistance amplifiers, is ignored in the backpropagation. The experimental success of the network indicates that such nonidealities are compensated for in the adaptive nature of the training algorithm.

The system of Fig. 1(b) functions well as a three-layer perceptron network. Training in the hardware network is comparable to that of the ideal (theoretical) network, and once trained, the hardware network may be operated as a standalone processor, with a processing time of -100 µs through the three-layer perceptron network.

In the system architecture of Fig. 1(b), input to the network occurs along the horizontal lines 14 of the photoconductor array 12, as a series of applied voltages. For many applications this may be appropriate. As discussed previously, one possible exception may occur in the case of image processing, which represents an important class of problems in the study of neural networks. For many image processing tasks it may be more desirable to create a compact network architecture that is able to sense an incident optical image and process the information directly. Because the present network is based upon the use of a photosensitive array, it is natural to investigate the extension of the system to allow optical image capture.

Turning next to Fig. 2(a), shown is an architecture of an optically-addressed neural network 40 used to directly allow optical imaging to be incorporated into the hardware neural network. A new region wji0 of transmissivity in the LCD array 42 is introduced into the system, through which is sensed an optical image projected onto it in window region wji0. Therefore, in this system, an incident image is confined to the upper left quadrant of the sensor array, suitably covering a region of 64 x 64 sensor nodes. Both of the weight fields wkjI and wlkII still exist, although their location on the LCD 42 is changed, as indicated in the figure. Consideration of the architecture reveals that the photoconductor array 12 essentially compresses the incident 64x64 pixel optical image into 32 lines of data (the 32 vertical line pairs illuminated by the incident image). These 32 lines are automatically routed into the horizontal lines of the weight field wkjI. The vertical lines of the weight field wkjI correspond to the hidden-layer neuron inputs, and their outputs are routed once again to the horizontal lines of weight field wlkII. The output of the network is as usual associated with the vertical lines of the weight field wlkII and is provided to amplifier 44. The amplifier 44 provides separate amplification to each signal line connected thereto.

Fig. 2(a) depicts only the column (vertical) lines being connected to amplifier (set) 44. It is to be appreciated that amplification may be alternatively provided, in whole, or in part, to selected horizontal (row) conductors. The dominant constraint is that when an impressed voltage is made on one row or column conductor, amplification must be associated with the complementary conductor of such node.

Two important points may be made immediately concerning the system of Fig. 2(a). First, the system allows the input of an image that is much larger (in terms of number of sensed pixels, here 64 x 64 = 4096) than the maximum input size of the neural network itself. This is a useful feature: it is very typical that two-dimensional images contain many more pixels than is convenient for direct input into a neural network, and hence some preprocessing transformation of the direct image is desirable (e.g. feature extraction). In the present system, the image is integrated along vertical lines, corresponding to the summing of currents in the photoconductor array. It is appreciated that for some applications, this particular method of compressing the image information may not be optimal. The issue will be discussed further below

A second important point concerning the system of Fig. 2(a) is that the LCD array 42 may be used as a programmable filter for the imaging process. Thus the LCD may be used to specifically tailor the transformation of the input image, in a manner that is suited for a given processing task. For example, one may program the LCD over the region wji0 to be sensitive to different spatial frequencies along different vertical lines of the incident image, by impressing upon the LCD patterns of appropriate spatial frequency modulation. More generally, as will be discussed below, the standard backpropagation training algorithm may be extended to allow adjustment of the weight field wji0. This allows the three weights fields wji0, wkjI, and wlkII to simultaneously be trained to perform a given image processing task.

In the system of Fig. 2(a), the weight fields wkjI and wlkII correspond to the standard interconnection weights of a three-layer perceptron network, and operate in a manner entirely comparable to the original architecture of Fig. 1(b). In particular, the LCD array 42 is uniformly illuminated over the areas corresponding to these weight fields, with incident light intensity I0. As discussed in Section 2, for the value of I0 typically used in the system, the range of these weight values may be taken to correspond to ± 1.0, when mapped to a theoretical network of tanh(x) transfer function.

The image to be processed is projected onto the LCD over the region corresponding to the weight field wji0. In this area, the incident light intensity varies spatially, and hence the underlying photoconductive sensors are affected by the product of the local LCD transmissivity and the incident illumination. As illustrated in FIG. 2(b), each weight wji0 corresponds to two photoconductive sensors, due to the differential pairing of vertical lines on the array, when input to nonlinear transresistance amplifiers 44 of the system. A representative weight wji0 is indicated in Fig. 2(b), with two interdigitated sensors each masked by 16 LCD pixels. The integrated transmissivities of the 16 LCD pixels above the two sensors are indicated as wji0+ and wji0- in Fig. 2(b). The values wji0+ and wji0- may be thought of as the weights that would be programmed if the incident illumination upon the region in question were equal to I0 (in which case wji0 = wji0+ - wji0-). The actual incident illumination is labelled lji+ and lji-, where these values are normalized relative to I0 (hence for an uniform illumination of l0, lji+ = lji- = 1.0). With these definitions, we may write: wji0 = wji0 + lji+ - wji0 - lji- Each partial weight wji0+ and wji0- is now governed by a 4 x 4 pixel group on the LCD, which is driven in the same quasi-halftone manner as described in Section 2, so that each partial weight may be programmed to one of 65 levels, corresponding to values between 0 and 1.0 in a theoretical network of tanh(x) neuron transfer function.

In the modified network of Fig. 2(a) there are still 'inputs' associated with voltages that drive the top 64 horizontal lines of the photoconductor array. In theory, these voltage levels may be included in the training process, though they remain static as different images are projected onto the LCD array 42. Upon examination of the architecture it is clear that any effect produced upon the system through variation of these 'input' voltage levels may be achieved alternatively by modification of the weight field wji0, with all 'input' levels reset to 1.0 (i.e. 10V on the sensor array) in this region. It is therefore much simpler to set all of the inputs to the weight field wji0 to a value of 1.0, which corresponds to programming the top 64 horizontal lines of the sensor array to a value of 10 volts. Note that in the hardware system, this means that one 10V power supply is able to drive the photoconductor array, greatly simplifying its operation.

The (idealized) forward propagation through the network of Fig. 2(a) therefore may be written as: fj = tanh ( Σi [wji0+ lji+ - wji0- lji-]) hk = tanh(Σj wkjl fj) ol = tanh ( Σk wlkll hk) where fj corresponds to the compressed image information, and is the input to the subsequent three-layer perceptron network. The quantities hk and ol correspond respectively to the hidden-layer and output-layer neuron values. The weights wji0+ and wji0- are constrained to lie within the range zero to one, and weights wkjI and wlkII are constrained to lie within the range ± 1.0.

The standard backpropagation training algorithm may be extended to the architecture of Fig. 2(a), by appropriate chain rule differentiation to calculate the values ∂E/∂Wji0+ and ∂E/∂Wji0-, where E is the network error, E = SΣl (tl-ol)2 and tl is the target value for output neuron ol. The resulting equations for updating the weights are: δwlkII(t) = vδwlkII(t-1) + (η/epoch_no) Σepoch hk(tl-to)(1-ol2), δwkjI(t) = vδwkjI(t-1) + (η/epoch_no) Σepoch fj (1-hk2) ΣlwlkII(tl-to)(1-ol2), δwji0γ(t) = vδwji0γ(t-1) + (η/epoch_no) Σepoch ljiγ(1-fj2) Σk,l wlkIIwkjI(tl-to) (1-ol2)(1-hk2). Here the learning coefficient is denoted by η and the momentum coefficient by v. The parameter γ in Eq. 4(c) represents either a + or - symbol. Herein, one epoch always corresponds to the presentation often input images (i.e. epoch.no = 10).

The network has been tested on two problems which form suitable examples: recognition of handprinted digits and recognition of human faces. In running the network, input images are projected onto the upper left quadrant of the LCD (see Fig. 2a), using a Sharp #XV-101TU LCD Video projector, whose video input is obtained from a video camera. The same projector is used to illuminate uniformly the weight fields wkjI and wlkII on the LCD. For both recognition tasks, the network used contains ten hidden units and five output units (thus in Fig. 2a, I = 64, J = 32, K = 10, and L = 5).

In training the network, exemplar images are not projected onto the LCD using the video projector. Instead, the training images are downloaded onto the LCD directly over the weight field wji0. Thus, for example, over a 4 x 4 pixel area of the LCD whose transmissivity would be programmed to a value wji0+ in the trained network, the transmissivity is programmed to the value lji + wji0+ during training. In the training phase the incident illumination over the entire LCD is uniform. This approach of imprinting the training images onto the LCD is assumed largely for convenience. Training images are stored digitally, as 64 x 64 pixel grayscale bitmap images. It is convenient to write these images onto the LCD during training, rather than feed them into the video projector. It would certainly be possible to use projected images for training the network, and this may yield better results in certain circumstances, as that is the mode in which the network is run, once trained.

For the problem of handprinted character recognition, the network is trained to recognize the digits 0 to 5. Because ultimately the network is expected to recognize direct, real-time images, it is not reasonable to rigorously normalize the characters to a bounding box, which is often done in such recognition tasks. It is clear however that due to the physical size limitation of the present hardware network (there are only ten hidden units), some constraint must be used in printing the digits. In creating a training set, and in testing the trained network, a peripheral box is employed inside of which the digits are printed. The digits are rendered so that they fill approximately the full height of the box (the width is not constrained). Fig. 3 shows examples of 10 training images.

In the structure exemplified by Fig. 2(a), realized is an optical neural network for a standard three-layer perceptron. The interconnection are illustrated in the node map provided Fig. 2(c).

Figs. 2(d) and 2(e) show the architecture and node map associated with a further interconnection realizable by a slight variation in mask construction, i.e., connections from layer K to layer J. It will be appreciated that the system differs from that in Fig. 2(a) by mask/interconnections selection. Similar pairings are provided by Figs. 2(f) and 2(g) (recurrent connection within layer K), Figs. 2(h) and 2(i) (recurrent connections within layer J), and Figs. 2(j) and 2(k) (connections from layer J to layer L). From these illustrations, it will be appreciated that many different neural interconnections are realizable by merely selecting the mask appropriately.

The illustrative network is trained on 300 exemplars, with weight fields wji0+ and wji0- initialized to random values in the range 0 to 0.3, and weights wkjI and wlkII randomly initialized to values in the range ± 0.2. During training, as well as running of the network, the top 64 horizontal lines of the photoconductor array are maintained at 10.0 V. One output neuron is assigned to each of the five digits, and is trained to produce a value of 0.8 (i.e. 8.0 V) if the output neuron corresponds to that digit, and a value of -0.8 (i.e. -8.0 V) otherwise. In Fig. 4 are shown data of the network error during training. Once trained, the system is exercised by projecting real-time images captured using the camera/video projector combination. In Fig. 5 are shown ten test images; the network output is indicated by histograms, with the five bins corresponding to the five output neuron values. It should be emphasized that the network response to the digits of Fig. 5 is performed in real time; the bitmap images shown in Fig. 5 were obtained by electronic scanning of the paper upon which they were written, after testing of the network.

The digits and histograms of Figs. 5(a) through 5(e) indicate the typical classification performance of the trained network. In Figs. 5(f) through 5(h) are shown sequential images of the creation of the digit four. The first vertical stroke in rendering the digit is recognized as a 1 (Fig. 5f). After the horizontal stroke is added (Fig. 5g), the network no longer classifies the image strongly as any of the five digits. With introduction of the final diagonal stroke (Fig. 5h), the network classifies the image as that of the digit four. Note that this rendering of a four is different than that shown in Fig. 5(e): the network was trained to recognize both cases. In Fig. 5(i) an example is shown of correct classification even when the digit does not touch both the top and bottom of the bounding box. Finally, Fig. 5(j) represents a case in which the network is unable to classify the image. It is interesting to note that the trained network performed best when the bounding box was included during testing. It might have been expected that since in training the bounding box was common to all exemplars the network would learn to ignore its presence, but this was not found to be the case.

Similarly, the network is trainable to recognize 64 x 64 pixel images of four different faces. In Fig. 6 are shown examples of eight training images. The network was trained on 250 exemplars, which included images of the faces under varying illumination, varying angle of lateral head tilt (within a lateral range of ± 15 degrees from direct view), varying magnification of the face (±5%), varying translation of the head within the frame ( ± 10%), and varying facial expression. To successfully train the network in facial recognition, it is desirable to begin with a structured weight field wji0, rather than the initial random pattern that sufficed for training of the network on handprinted digits. The initial weight field wji0 used for facial recognition is indicated in Fig. 7. The pattern of the weight field is seen to select different spatial frequencies of the image, along its vertical columns. Every four vertical line pairs of the sensor array (i.e. 8 lines) perform a crude spatial-frequency compression (in the vertical dimension) of the corresponding portion of the image. With this initial weight field wji0, and random values for the weights wkjI and wlkII, the network is successfully trainable. A typical training error curve is shown in Fig. 8. In Fig. 9 are shown ten examples of the response of the trained network to real-time video images of the four faces.

In Figures 9(a) through 9(e) and 9(f) through 9(i) are shown examples of correct network classification. The figures represent notable variation in image capture: there is significant lateral tilting of the head in Figs. 9(f), (g), and (i). Furthermore, the face of Fig. 9(f) is translated within the frame of the image. The illumination has been altered in generating the images of Figs. 9(b) and 9(h). Examples of poor network classification are shown in Figs. 9(e) and 9(j). The illumination of the head in Fig. 9(j) may be too different from that in the training exemplars for the network to correctly classify the image. The reason for incorrect classification of the image in Fig. 9(e) is not clear, but may be associated with an unacceptable reduction of the head size. These results on recognition of faces may be compared to recent work by others, using photorefractive holograms.

The disclosed network is shown to train well, considering its size, in recognition of digit and facial images. It may be noted that the results of Figs. 4, 5, 8, and 9 are found to be very similar to simulated results of an ideal network, using Equations (1) - (4) to describe forward propagation and training. In particular, after 750 epochs, the training error in the simulated network is only -30% below that of the hardware network, for both the digit and facial recognition tasks. Larger networks are currently being simulated, to understand the capability of the system.

The above results indicate the opticallyaddressed neural network architecture to allow direct projection of optical images onto the network, with subsequent neural network processing of the sensed images. This may be a very attractive and compact architecture for some image recognition tasks. The network, once trained, responds in real time to images projected onto it. The response time of the trained network corresponds to the combined response times of the photoconductive sensors and the nonlinear transresistance amplifier circuitry. The present transresistance amplifier circuitry has a response time of 40µs. In previous work, it has been shown that the response time of the sensors to changes in incident illumination is of the order of 200 - 300 µs. This is much longer than the response time of the sensors to a change in applied voltage at constant illumination (several microseconds), and hence is expected to limit the processing speed of the trained system.

The training time of the network is limited by two factors. First, there is the time needed to measure the hardware neuron values, and perform the subsequent backpropagation of the network error in the digital computer. Second, there is the time associated with changing the pattern on the LCD array, either merely updating the interconnection weight fields, or additionally impressing the training image onto the upper-left quadrant of the LCD. In the present implementation, training occurs at a rate of 0.3 epoch/sec. (3 exemplars/sec.).

It is useful to bring to attention the areas in the network over which the LCD pixels are maintained in a state of minimum transmittance (the black regions shown in the schematics of Figs. 1b and 2a). These areas typically correspond to interconnections not used in the present multilayer feedforward network, such as recurrent connections within layers, or connections between non-sequential layers (e.g. between input and output layers). The ability to implement such interconnections occurs naturally in the architecture of the present system. For standard feedforward networks, this results in some inefficiency in implementation, as a significant fraction of the possible network interconnections are left unused. It is possible to implement the present feedforward architecture more efficiently by driving and sensing the horizontal and vertical lines from all four sides of the sensor array. In Fig. 10 is shown an example of such an efficient implementation. Here the conductive rows 50 and columns 52 of the sensor array 54 are severed (electrically isolated) along the two dotted lines shown. An optical image is incident upon the upper half of the sensor array and super-imposed LCD 60. The system indicated in Fig. 10 allows (64-L) hidden layer unit's with L outputs. For such a network, only 2L2 sensors are not utilized. Photoconductive outputs are committed to amplifiers 62 and 64. This represents a very efficient use of available interconnections.

As mentioned earlier, the image-sensing architecture of the present network may be optimized further. In the disclosed embodiment, incident optical image is integrated along vertical lines 52, due to the hardwired interconnection of the photoconductive sensor array. This sensing architecture is capable of producing successful pattern recognition. Improved network performance may be expected if the incident optical image were transformed differently upon its initial capture. In particular, rather than integrate throughout entire rows or columns of the image, it may be more suitable to capture and process local regions of the image, passing this local information on to the following layer of the network. In the region of the photoconductive sensor array dedicated to image capture, photoconductive structures (or alternatively photodiode structures) is suitably fabricated which are each sensitive to a local area of the incident image. These structures could be made to allow bipolar weighting of the incident image, if desired. Furthermore, retaining the LCD as a spatial light modulator over this region would allow the system to perform adaptive local filtering of the optical image. This approach would be more consistent with other neural network architectures which employ local feature extraction as an initial step in image processing.

Further adaptation includes the use of a lenslet array to project duplicate images onto the sensor array. In this case, any given region of the optical image would be processed by multiple sensors, allowing multiple features to be extracted for each region of the image.

In the preferred embodiment, the nonlinear current-to-voltage converters used to perform sigmoidal transformation are being integrated onto the glass substrate of the photoconductive array, using polycrystalline silicon technology. Successful integration of these amplifiers should allow the entire system to be contained within the photoconductor array substrate and LCD. In particular, only five external voltage lines will be needed to drive the entire sensor array and amplification circuitry. In addition, if after training the network to perform a specific task, the LCD were replaced by an appropriate static spatial light modulator (e.g photographic film), the entire neural network module would require only these few input voltage lines to capture and process an incident optical image.

A hardware neural network architecture has been taught, which is capable of capturing and processing an incident optical image, in real time. The system, based on the combination of a photoconductive array and LCD, operates in a standalone mode, once trained. This architecture allows the filtering of the optical image upon capture to be incorporated into the network training process. The network has the potential to be very compact. Because all of the network components are based upon large-area thin-film technology, there is great potential for scalability and integration within the architecture.


Anspruch[de]
  1. Optisch adressiertes neuronales Netz (40) mit:
    • einer zweidimensionalen Matrix (12) aus Photoleitern, die eine Matrix aus M mal N Matrixknoten (18) definieren, wobei M und N durch positive Ganzzahlen definiert werden und die Knoten jeder Spalte einen gemeinsamen Spaltenleiter (16) verwenden und die Knoten jeder Reihe einen gemeinsamen Reihenleiter (14) verwenden, um synaptische Verknüpfungen zu bilden,
    • einer Lichtmaske (42), die zwischen der zweidimensionalen Matrix und einer assoziierten Lichtquelle angeordnet ist, so dass ein allgemein kollimiertes Licht aus derselben auf einen ausgewählten Teil der Vielzahl von Matrixknoten fällt, wobei die lokale Durchlässigkeit der Lichtmaske gewählt werden kann, um den Widerstand der Photoleiter zu steuem,
    • einem Eingangsteil der Matrixphotoleiter, der ein Teilsatz der zweidimensionalen Matrix von Photoleitem ist, wobei jeder gemeinsame Reihenleiter des Eingangsteils eine Spannung aufweist, die auf einen vorgewählten Pegel festgelegt ist,
    • einer Vielzahl von Verstärkern (44), die jeweils operativ an ihrem Eingang mit einem ausgewählten der gemeinsamen Spaltenleiter (16) der Matrix verbunden sind,
    • einer ersten Vielzahl von Leitern zum Aufprägen von ausgewählten verstärkten Signalen, die von einer ersten Gruppe von Verstärkern aus der Vielzahl von Verstärkern empfangen werden, auf die Reihenleiter eines ausgewählten Rückkopplungsteils der zweidimensionalen Matrix, wobei der Rückkopplungsteil der Matrix keine Reihenleiter gemeinsam mit dem Eingangsteil verwendet, und
    • einer zweiten Vielzahl von Leitern, die mit einer zweiten Gruppe von Verstärkern assoziiert sind, die eine Systemausgabe aus ausgewählten verstärkten Signalen bilden, die von den Verstärkern der zweiten Gruppe empfangen werden,
    gekennzeichnet durch
    • Maskendurchlässigkeiten, die mit dem Eingangsteil assoziiert und derart angeordnet sind, dass sie durch eine Rückwärtsfortpflanzung trainiert werden können, und
    • eine Bildprojektionseinrichtung zum Projizieren eines optischen Bildes durch die Lichtmaske (42) auf den Eingangsteil der Photoleiter.
  2. Optisch adressiertes neuronales Netz nach Anspruch 1, wobei die zweite Vielzahl von Leitern keinen Verstärker gemeinsam verwenden, der mit einem gemeinsamen Spaltenleiter des rechteckigen Eingangsteils assoziiert ist.
  3. Optisch adressiertes neuronales Netz nach Anspruch 2, wobei die Verstärker nichtlineare Strom-zu-Spannungs-Wandler umfassen.
  4. Optisch adressiertes neuronales Netz nach Anspruch 3, wobei die Lichtmaske das allgemein kollimierte Licht der assoziierten Lichtquelle auf einen ersten Teilsatz der Matrixelemente richtet, die derart definiert sind, dass sie keinen gemeinsamen Reihenleiter mit dem rechteckigen Eingangsteil verwenden.
  5. Optisch adressiertes neuronales Netz nach Anspruch 4, wobei die Lichtmaske das allgemein kollimierte Licht der assoziierten Lichtquelle auf den ersten Teilsatz der Matrixelemente richtet, die derart definiert sind, dass sie wenigstens einen Teil von gemeinsamen Reihenleitern verwenden, die mit dem Systemausgang assoziiert sind.
  6. Optisch adressiertes neuronales Netz nach Anspruch 5, wobei der erste Teilsatz rechteckig ist.
  7. Optisch adressiertes neuronales Netz nach Anspruch 5 oder 6, wobei der erste Teilsatz von einem ersten und einem zweiten rechteckigen Teilbereich umfasst wird, die dazwischen keinen gemeinsamen Reihenleiter verwenden.
  8. Optisch adressiertes neuronales Netz nach Anspruch 7, wobei der erste und der zweite rechteckige Teilbereich dazwischen keinen gemeinsamen Spaltenleiter verwenden.
  9. Optisch adressiertes neuronales Netz mit:
    • einer zweidimensionalen Matrix (12) mit Photoleitern mit zwei Anschlüssen in einer Matrix aus M Reihen mal N Spalten, wobei M und N als positive Ganzzahlen definiert sind,
    wobei
    • ein erster Anschluss jedes Photoleiters in jeder Reihe M von N Photoleitern mit einem M-Reihenleiter gemeinsam verbunden ist, und
    • ein zweiter Anschluss jedes Photoleiters in jeder Spalte N von M Photoleitem mit einem N-Spaltenleiter gemeinsam verbunden ist,
    • einer zweidimensionalen Maske (42) mit einer lokal auswählbaren Durchlässigkeit, die allgemein parallel zu der Anordnung zwischen der Anordnung und einer assoziierten und relativ kollimierten Lichtquelle vorgesehen ist, wobei die Maske eine Vielzahl von ausgewählten Teilbereichen umfasst, die auf eine ausgewählte Durchlässigkeitsstufe gesetzt sind,
    • einer Vielzahl von J benachbarten Eingangsleitem, die jeweils elektrisch mit einem entsprechenden Reihenleiter verbunden sind, wobei J eine Ganzzahl kleiner M ist,
    • einer Vielzahl von Verstärkern (44), die elektrisch mit jedem Spaltenleiter verbunden sind, so dass N verstärkte Spaltensignale erzeugt werden,
    • einer elektrischen Verbindung, die jeden ersten Teilsatz der N verstärkten Spaltensignale mit einem einzigen der J benachbarten Eingangsleiter verbindet, und
    • einem Ausgang, der aus einem zweiten Teilsatz der N verstärkten Spaltensignale gebildet wird,
    wobei die ausgewählten Teilbereiche der Maske umfassen:
    • einen ersten Fensterteil, der einen Lichtdurchgang festlegt, um so einen Teilsatz wkj der Photoleiter zu beeinflussen, wobei der Teilsatz durch eine Anzahl von k Reihen von J Spalten der Photoleiter definiert wird, und
    • einen zweiten Fensterteil, der einen Lichtdurchgang festlegt, um so einen Teilsatz wlk der Photoleiter zu beeinflussen, wobei der Teilsatz durch eine Anzahl von k Spalten und I Reihen der Photoleiter definiert wird, wobei der erste und der zweite Teilsatz keine gemeinsamen Reihen oder Spalten aufweisen,
    gekennzeichnet durch
    • einen dritten Fensterteil, der das optische Bild zu einem Bereich wji der Photoleiter durchlässt, wobei der Bereich keine Photoleiter der Teilsätze wkj oder wlk gemeinsam verwendet und wobei die mit den Photoleitern des Bereichs wji assoziierten Spaltenleiter mit dem ersten Teilsatz der N verstärkten Spaltensignale assoziiert sind, die mit den J Eingangsleitern verbunden sind, und wobei die mit dem Bereich wji der Photoleiter assoziierten Reihenleiter auf einer festgelegten Spannung gehalten werden, wobei der dritte Fensterteil der Maske in Reaktion auf Signale aus dem Ausgang gesteuert werden kann.
  10. Neuronales Verfahren mit folgenden Schritten:
    • Vorsehen einer zweidimensionalen Matrix (12) aus Photoleitern, die eine Matrix aus M mal N Matrixknoten (18) definieren, wobei M und N durch positive Ganzzahlen definiert werden, die Knoten jeder Spalte einen gemeinsamen Spaltenleiter (16) verwenden und die Knoten jeder Reihe einen gemeinsamen Reihenleiter (14) verwenden, um synaptische Verknüpfungen zu bilden,
    • Vorsehen einer Lichtmaske (42), die zwischen der zweidimensionalen Matrix und einer assoziierten Lichtquelle vorgesehen ist, so dass ein allgemein kollimiertes Licht aus derselben auf einen ausgewählten Teil der Vielzahl von Matrixknoten fällt, wobei die lokale Durchlässigkeit der Lichtmaske ausgewählt werden kann, um den Widerstand der Photoleiter zu steuern,
    • Auswählen eines Eingangsteils der Matrixphotoleiter als einen Teilsatz der zweidimensionalen Matrix von Photoleitern, wobei jeder gemeinsame Reihenleiter des Eingangsteils eine Spannung aufweist, die auf einem vorgewählten Pegel festgelegt ist,
    • Vorsehen einer Vielzahl von Verstärkern (44), die jeweils operativ an ihrem Eingang mit einem ausgewählten gemeinsamen Spaltenleiter (16) der Matrix verbunden sind,
    • Vorsehen einer ersten Vielzahl von Leitern für das Aufprägen von ausgewählten verstärkten Signalen, die von einer ersten Gruppe von Verstärkern aus der Vielzahl von Verstärkern empfangen werden, auf die Reihenleiter eines Rückkopplungsteils der zweidimensionalen Matrix, wobei der Rückkopplungsteil der Matrix keine gemeinsamen Reihenleiter mit dem Eingangsteil verwendet, und
    • Vorsehen einer zweiten Vielzahl von Leitern, die mit einer zweiten Gruppe von Verstärkern assoziiert sind, um eine Systemausgabe aus ausgewählten verstärkten Signalen zu bilden, die aus den Verstärkern der zweiten Gruppe empfangen werden,
    • Vorsehen eines Rückwärtsfortpflanzungsalgorithmus zum Trainieren von Maskendurchlässigkeiten, die mit dem Eingangsteil assoziiert sind, und
    • Vorsehen einer Bildprojektionseinrichtung zum Projizieren eines optischen Bildes durch die Lichtmaske (42) auf den Eingangsteil der Photoleiter,
    • Übertragen eines optischen Bildes über die Lichtmaske zu dem ausgewählten Eingangsbereich der zweidimensionalen Matrix, wobei das optische Bild in das trainierbare neuronale Netz eingegeben wird,
    • Verstärken der Signale der gemeinsamen Spaltenleiter, und
    • Aufprägen von wenigstens einigen der verstärkten Signale, die von den Verstärken empfangen werden, auf einen Matrixrückkopplungsteil der Matrixknoten, und Erzeugen eines Ausgangssignals, das als Ergebnis des empfangenen optischen Bildes gebildet wird.
Anspruch[en]
  1. An optically-addressed neural network (40) comprising:
    • a two-dimensional array (12) of photoconductors defining an array of M by N array nodes (18), wherein M and N are defined by positive integers, and nodes of each column share a common column conductor (16) and nodes of each row share a common row conductor (14), thereby forming synaptic interconnections;
    • a light mask (42) disposed between the two-dimensional array and an associated light source such that generally collimated light therefrom is incident on a selected portion of the plurality of array nodes, the local translucence of the light mask being selectable so as to control the resistance of the photoconductors;
    • an input portion of the array photoconductors being a subset of the two-dimensional array of photoconductors, each common row conductor of the input portion having a voltage thereon fixed at a preselected level;
    • a plurality of amplifiers (44) each operatively connected at their input to a selected one of the common column conductors (16) of the array;
    • a first plurality of conductors for impressing selected amplified signals, received from a first group of amplifiers among the plurality of amplifiers, to the row conductors of a selected feedback portion of the two-dimensional array, said feedback portion of the array not sharing any row conductors with said input portion; and
    • a second plurality of conductors, associated to a second group of amplifiers forming a system output from selected amplified signals received from the amplifiers of the second group;
    characterized by
    • mask translucencies associated with the input portion being arranged to be trainable by back propagation; and
    • image projecting means for projecting an optical image through said light mask (42) onto said input portion of said photoconductors.
  2. The optically-addressed neural network of claim 1, wherein the second plurality of conductors share no amplifier associated with any common column conductor of the rectangular input portion.
  3. The optically-addressed neural network of claim 2, wherein the amplifiers are comprised of nonlinear current-to-voltage converters.
  4. The optically-addressed neural network of claim 3, wherein the light mask directs the generally collimated light of the associated light source to a first subset of the array elements defined as sharing no common row conductor with the rectangular input portion.
  5. The optically-addressed neural network of claim 4, wherein the light mask directs the generally collimated light of the associated light source to the first subset of the array elements defined as sharing at least a portion of common row conductors associated with the system output.
  6. The optically-addressed neural network of claim 5, wherein the first subset is rectangular.
  7. The optically-addressed neural network of claims 5 or 6, wherein the first subset is comprised of first and second rectangular sub-portions sharing no common row conductor therebetween.
  8. The optically-addressed neural network of claim 7, wherein first and second rectangular sub-portions share no common column conductor therebetween.
  9. An optically-addressed neural network comprising:
    • a two-dimensional array (12) of two-terminal photoconductors disposed in an array of M rows by N columns, wherein M and N are defined as positive integers, wherein
    • a first terminal of each photoconductor in each row M of N photoconductors is connected in common to a row M conductor; and
    • a second terminal of each photoconductor in each column N of M photoconductors is connected in common to a column N conductor;
    • a two-dimensional mask (42) of locally selectable translucence, generally parallel to the array, disposed between the array and an associated, relatively collimated light source, the mask including a plurality of selected sub-areas each being set to a selected translucence level;
    • a plurality of J adjacent input conductors, each being electrically coupled to a corresponding one of the row conductors, wherein J is an integer less than M;
    • a plurality of amplifiers (44) electrically coupled to each column conductor such that N amplified column signals are generated thereby;
    • an electrical coupling electrically connecting each of a first subset of the N amplified column signals to a single one of said J adjacent input conductors; and
    • an output formed from a second subset of the N amplified column signals;
    • the selected sub-areas of the mask including:
      • a first window portion defining light passing therethrough so as to affect a subset wkj of the photoconductors, which subset is defined by a subset of a number k rows and J columns of the photoconductors, and
      • a second window portion defining light passing therethrough so as to affect a subset wlk of the photoconductors, which subset is defined by a subset of k columns and I rows of the photoconductors, the first and second subset having neither rows nor columns in common,
      characterized by
    • a third window portion so as to communicate the optical image to an area wji of the photoconductors, which area shares no photoconductors of either subset wkj or wlk, and wherein column conductors associated with photoconductors of the area wji are associated to the first subset of N amplified column signals coupled to the J input conductors, and wherein said row conductors associated to said area wji of said photoconductors are held at a fixed voltage, said third window portion of said mask being controllable in response to signals from said output.
  10. A neural method comprising the steps of:
    • providing a two-dimensional array (12) of photoconductors defining an array of M by N array nodes (18), wherein M and N are defined by positive integers, and nodes of each column share a common column conductor (16) and nodes of each row share a common row conductor (14), thereby forming synaptic interconnections;
    • providing a light mask (42) disposed between the two-dimensional array and an associated light source such that generally collimated light therefrom is incident on a selected portion of the plurality of array nodes, the local translucence of the light mask being selectable so as to control the resistance of the photoconductors;
    • selecting an input portion of the array photoconductors being a subset of the two-dimensional array of photoconductors, each common row conductor of the input portion having a voltage thereon fixed at a preselected level;
    • providing a plurality of amplifiers (44) each operatively connected at their input to a selected one of the common column conductors (16) of the array;
    • providing a first plurality of conductors for impressing selected amplified signals, received from a first group of amplifiers among the plurality of amplifiers, to the row conductors of a feedback portion of the two-dimensional array, said feedback portion of the array not sharing any row conductors with said input portion; and
    • providing a second plurality of conductors, associated to a second group of amplifiers forming a system output from selected amplified signals received from the amplifiers of the second group;
    • providing a back propagation algorithm for training mask translucencies associated with the input portion; and
    • providing image projecting means for projecting an optical image through said light mask (42) onto said input portion of said photoconductors,
    • transmitting an optical image, via the light mask to the selected input area of the two-dimensional array, whereby the optical image is input into the trainable neural network,
    • amplifying the signals of the common column conductors; and
    • impressing at least some of the amplified signals received from the amplifiers, to an array feedback portion of the array nodes; and generating an output signal formed as a result of the received optical image.
Anspruch[fr]
  1. Réseau neuronal adressé optiquement (40) comprenant :
    • une matrice à deux dimensions (12) de photoconducteurs définissant une matrice de M sur N noeuds de matrice (18), où M et N sont définis par des nombres entiers positifs, et des noeuds de chaque colonne partagent un conducteur de colonne commun (16) et des noeuds de chaque rangée partagent un conducteur de rangée commun (14), en formant ainsi des interconnexions synaptiques,
    • un masque de lumière (42) disposé entre la matrice à deux dimensions et une source de lumière associée de manière à ce que de la lumière généralement collimatée provenant de celle-ci soit incidente sur une partie sélectionnée de la pluralité de noeuds de la matrice, la translucidité locale du masque de lumière pouvant être sélectionnée de façon à commander la résistance des photoconducteurs,
    • une partie d'entrée des photoconducteurs de la matrice constituant un sous-ensemble de la matrice à deux dimensions de photoconducteurs, chaque conducteur de rangée commun de la partie d'entrée présentant une tension sur celui-ci fixée à un niveau présélectionné,
    • une pluralité d'amplificateurs (44) reliés chacun de façon fonctionnelle à leur entrée, à un conducteur sélectionné parmi les conducteurs de colonnes communs (16) de la matrice,
    • une première pluralité de conducteurs destinés à appliquer des signaux amplifiés sélectionnés, reçus d'un premier groupe d'amplificateurs parmi la pluralité d'amplificateurs, aux conducteurs de rangées d'une partie de contre-réaction sélectionnée de la matrice à deux dimensions, ladite partie de contre-réaction de la matrice ne partageant aucun conducteur de rangée avec ladite partie d'entrée, et
    • une seconde pluralité de conducteurs, associés à un second groupe d'amplificateurs formant une sortie de système parmi les signaux amplifiés sélectionnés reçus des amplificateurs du second groupe,
       caractérisé par
    • des translucidités de masque associées à la partie d'entrée qui sont agencées pour pouvoir subir un apprentissage par une contre-propagation, et
    • un moyen de projection d'image destiné à projeter une image optique au travers dudit masque de lumière (42) sur ladite partie d'entrée desdits photoconducteurs.
  2. Réseau neuronal adressé optiquement selon la revendication 1, dans lequel la seconde pluralité de conducteurs ne partagent aucun amplificateur associé à un conducteur de colonne commun quelconque de la partie d'entrée rectangulaire.
  3. Réseau neuronal adressé optiquement selon la revendication 2, dans lequel les amplificateurs sont constitués de convertisseurs de courant en tension non linéaires.
  4. Réseau neuronal adressé optiquement selon la revendication 3, dans lequel le masque de lumière oriente la lumière généralement collimatée de la source de lumière associée vers un premier sous-ensemble des éléments de matrice définis comme ne partageant pas de conducteurs de rangée communs avec la partie d'entrée rectangulaire.
  5. Réseau neuronal adressé optiquement selon la revendication 4, dans lequel le masque de lumière oriente la lumière généralement collimatée de la source de lumière associée vers le premier sous-ensemble d'éléments de matrice définis comme partageant au moins une partie des conducteurs de rangées communs associée à la sortie du système.
  6. Réseau neuronal adressé optiquement selon la revendication 5, dans lequel le premier sous-ensemble est rectangulaire.
  7. Réseau neuronal adressé optiquement selon la revendication 5 ou 6, dans lequel le premier sous-ensemble est constitué de première et seconde sous-parties rectangulaires ne partageant pas de conducteurs de rangées communs entre celles-ci.
  8. Réseau neuronal adressé optiquement selon la revendication 7, dans lequel des première et seconde sous-parties rectangulaires ne partagent aucun conducteur de colonne commun entre celles-ci.
  9. Réseau neuronal adressé optiquement comprenant :
    • une matrice à deux dimensions (12) de photoconducteurs à deux bornes disposés suivant une matrice de M rangées sur N colonnes, où M et N sont définis comme des nombres entiers positifs, dans lequel,
    • une première borne de chaque photoconducteur dans chaque rangée M de N photoconducteurs est reliée en commun à un conducteur de rangée M, et
    • une seconde borne de chaque photoconducteur dans chaque colonne N de M photoconducteurs est reliée en commun à un conducteur de colonne N,
    • un masque à deux dimensions (42) dont la translucidité peut être sélectionnée localement, généralement parallèle à la matrice, disposé entre la matrice et une source de lumière relativement collimatée associée, le masque comprenant une pluralité de sous-zones sélectionnées, chacune étant établie à un niveau de translucidité sélectionnée,
    • une pluralité de J conducteurs d'entrée adjacents, chacun étant relié électriquement à un conducteur correspondant parmi les conducteurs de rangées, dans lequel J est un nombre entier inférieur à M,
    • une pluralité d'amplificateurs (44) reliés électriquement à chaque conducteur de colonne de sorte que N signaux de colonnes amplifiés sont générés de cette manière,
    • un couplage électrique reliant électriquement chaque signal d'un premier sous-ensemble des N signaux de colonnes amplifiés à un seul conducteur desdits J conducteurs d'entrée adjacents, et
    • une sortie formée d'un second sous-ensemble des N signaux de colonnes amplifiés,
    les sous-zones sélectionnées du masque comprenant :
    • une première partie de fenêtre définissant la lumière passant au travers de celle-ci de façon à affecter un sous-ensemble wkj des photoconducteurs, lequel sous-ensemble est défini par un sous-ensemble d'un nombre de k rangées et de J colonnes des conducteurs, et
    • une seconde partie de fenêtre définissant la lumière passant au travers de celle-ci de façon à affecter un sous-ensemble wlk des photoconducteurs, lequel sous-ensemble est défini par un sous-ensemble de k colonnes et 1 rangées des photoconducteurs, les premier et second sous-ensembles ne comportant ni rangées ni colonnes en commun,
       caractérisé par
    • une troisième partie de fenêtre de façon à communiquer l'image optique à une zone wji des photoconducteurs, laquelle zone ne partage aucun photoconducteur soit du sous-ensemble wkj, soit du sous-ensemble wlk, et dans lequel les conducteurs de colonnes associés aux photoconducteurs de la zone wji sont associés au premier sous-ensemble de N signaux de colonnes amplifiés couplés aux J conducteurs d'entrée, et dans lequel lesdits conducteurs de rangées associés à ladite zone wji desdits photoconducteurs sont maintenus à une tension fixe, ladite troisième partie de fenêtre dudit masque pouvant être commandée en réponse à des signaux provenant de ladite sortie.
  10. Procédé neuronal comprenant les étapes consistant à :
    • fournir une matrice à deux dimensions (12) de photoconducteurs définissant une matrice de M sur N noeuds de matrice (18), où M et N sont définis par des nombres entiers positifs, et des noeuds de chaque colonne partagent un conducteur de colonne commun (16) et des noeuds de chaque rangée partagent un conducteur de rangée commun (14), en formant ainsi des interconnexions synaptiques,
    • fournir un masque de lumière (42) disposé entre la matrice à deux dimensions et une source de lumière associée de sorte que de la lumière généralement collimatée de celle-ci soit incidente sur une partie sélectionnée de la pluralité de noeuds de matrice, la translucidité locale du masque de lumière pouvant être sélectionnée de façon à commander la résistance des photoconducteurs,
    • sélectionner une partie d'entrée des photoconducteurs de matrice qui constituent un sous-ensemble de la matrice à deux dimensions des photoconducteurs, chaque conducteur de rangée commun de la partie d'entrée présentant une tension de celui-ci fixée à un niveau présélectionné,
    • fournir une pluralité d'amplificateurs (44), chacun étant relié de façon fonctionnelle à son entrée à un conducteur sélectionné parmi les conducteurs de colonnes communs (16) de la matrice,
    • fournir une première pluralité de conducteurs destinés à appliquer des signaux amplifiés sélectionnés reçus d'un premier groupe d'amplificateurs parmi la pluralité d'amplificateurs, aux conducteurs de rangées d'une partie de contre-réaction de la matrice à deux dimensions, ladite partie de contre-réaction de la matrice ne partageant aucun conducteur de rangée avec ladite partie d'entrée, et
    • fournir une seconde pluralité de conducteurs associée à un second groupe d'amplificateurs formant une sortie de système à partir des signaux amplifiés sélectionnés reçus des amplificateurs du second groupe,
    • fournir un algorithme de contre-propagation destiné à l'apprentissage des translucidités de masque associées à la partie d'entrée, et
    • fournir un moyen de projection d'image destiné à projeter une image optique au travers dudit masque de lumière (42) sur ladite partie d'entrée desdits photoconducteurs,
    • transmettre une image optique, par l'intermédiaire du masque de lumière à la zone d'entrée sélectionnée de la matrice à deux dimensions, grâce à quoi l'image optique est appliquée en entrée dans le réseau neuronal pouvant subir un apprentissage,
    • amplifier les signaux des conducteurs de colonnes communs, et
    • appliquer au moins certains des signaux amplifiés reçus des amplificateurs, à une partie de contre-réaction de matrice des noeuds de la matrice, et générer un signal de sortie formé en tant que résultat de l'image optique reçue.






IPC
A Täglicher Lebensbedarf
B Arbeitsverfahren; Transportieren
C Chemie; Hüttenwesen
D Textilien; Papier
E Bauwesen; Erdbohren; Bergbau
F Maschinenbau; Beleuchtung; Heizung; Waffen; Sprengen
G Physik
H Elektrotechnik

Anmelder
Datum

Patentrecherche

Patent Zeichnungen (PDF)

Copyright © 2008 Patent-De Alle Rechte vorbehalten. eMail: info@patent-de.com