Optical neuron unit and network thereof

Abstract

The invention relates to an optical neuron unit and a network thereof. An artificial neuron network and corresponding neuron units and corresponding neuron units are described. The neuron network comprises a plurality of two-layer or more-layer artificial neuron units. Layers of artificial neuron units are configured to communicate therebetween via an arrangement of two or more optical waveguides (optical fibers). The arrangement of the two or more optical waveguides is configured to have a predetermined coupling between the two or more waveguides to provide cross-communication between the neuron units of the two or more layers.

Description

[0001] This application is a divisional application of an application with a filing date of September 30, 2020, application number 202080067486.4, and invention name “Optical neuron unit and its network”. Technical Field

[0002] The present invention relates to optical computing devices, and more particularly to optical computing devices and device configurations suitable for use in optically integrated artificial neuronal networks. Background Art

[0003] Optical computing uses the manipulation of visible or infrared light to perform computational processes, rather than using electric currents in electronic computing. In general, optical computing is able to achieve faster computational rates than electronic systems. This is in part because operations on pulses of light can occur faster and can allow for higher bandwidth information transmission. For example, current signals propagate at only about 10% of the speed of light due to the much larger dielectric constant in the microwave range relative to optical systems, which increases the computational rate of optical computing by almost a factor of 10. Because photons, unlike electrons, do not have polarity and do not carry charge, photons also consume less power and are less susceptible to cross-interference from neighboring electric fields.

[0004] Conventional optical processing systems typically utilize hybrid electrical-optical processing, often referred to as optoelectronic processing. In these systems, optical signals are used for data transmission and certain processing operations, and are converted into electrical signals for certain other processing operations. Such optoelectronic devices lose approximately 30% of their energy when converting electronic energy into photons and back. In addition, the conversion of optical signals to electrical signals and back slows down the transmission and processing of data. A large amount of research work is directed toward all-optical computing, which eliminates the need for optical-electrical-optical (OEO) conversion, thereby reducing the need for electrical energy and increasing processing rates.

[0005] Another favorable aspect of the field of optical computing is the implementation of artificial neural networks (ANNs). In general, neural network systems provide processing that can solve problems in a manner corresponding to the operation of the human brain. Artificial neural networks are basically computer systems inspired by biological neural networks (BNNs) that make up the brain. These systems "learn" to improve their performance in executing a set of commands to complete tasks of interest. More specifically, ANNs develop their set of relevant features from the learning material provided to it for optimizing the processing of relevant inputs about the selected task. Typical ANN systems are based on a collection of connected units or nodes called artificial neurons, which are artificial equivalents of biological neurons that make up the brain's BNNs. The connections between nodes, which are the artificial equivalents of biological synapses, can transmit signals from one node to another. The artificial neurons that receive the signal are configured to process the signal and then transmit the corresponding signal to the artificial neurons connected to it. Typically, artificial neurons are arranged in layers. Different layers can perform different kinds of transformations on their inputs and transmit corresponding output signals. The signal travels from the first layer (input layer) to the last layer (output layer), which may require several passes through different layers.

[0006] In previous work, the inventors of the present invention have developed neural network configurations that rely on light propagation and coupling between multi-core and multimode optical fibers.

[0007] For example, WO 2017 / 033197 by Zalevsky et al. teaches an integrated optical module. The optical module includes multiple optically coupled channels and can be used in an artificial neural network (ANN). According to some embodiments, the integrated optical module includes a multi-core optical fiber, wherein the cores are optically coupled.

[0008] WO 2019 / 186548 describes an artificial neuron unit and a neural network for processing input light. The artificial neuron unit includes a mode mixing unit (e.g., a multimode optical fiber) and a filtering unit, the mode mixing unit being configured to receive the input light and apply a selected mixing to light components of two or more modes within the input light, and provide an outgoing light, and the filtering unit being configured to apply a preselected filter to the outgoing light to select one or more modes of the outgoing light, thereby providing the output light of the artificial neuron unit. Summary of the invention

[0009] As described above, coupling of light propagating in optical fibers can be used for a variety of processing tasks suitable for use in neural network processing. However, there is a need in the art for a configuration of an operating all-optical neural network that can use optical manipulation to handle data transmission and processing. Typically, in conventional optical networks, optical components are relatively limited in handling nonlinear / processing operations. These functions are currently operated through electronic processing, the use of high-energy laser devices, and / or the operation of cold atoms. Each of these techniques has its disadvantages.

[0010] The present invention provides a processing solution suitable for implementation in an optical artificial neural network. The technology of the present invention is based on the use of multi-mode and multi-core optical fibers, and the use of the free space propagation properties of light to achieve the design of a fully operational all-optical neural network.

[0011] Therefore, the present invention provides an artificial neural network comprising a plurality of artificial neurons formed by optical waveguides. The prior art as described above with reference to WO 2017 / 033197 and WO 2019 / 186548 describes the use of multi-mode and multi-core 1D, 2D and 3D waveguides to provide integrated optical modules and artificial neuron units. The present technology further expands the constituent elements of the neural network architecture, providing controlled coupling, processing operations and training related processes, as described in more detail below.

[0012] As described in more detail below, the present technology provides an artificial optical neuron unit and a neuron network that allows additional operations of mixing, gain, and selection to be optically applied to signals within the neuron unit. Typically, an artificial neuron network is trained for a selected task by adjusting the weights of signal portions transmitted through different neurons and selected signal paths through the network. The present technology provides an optical arrangement and selected signal portion mixing that allows selected gain / pumping to be applied to spatial and / or temporal signal portions, and the selected signal portion mixing allows weight adjustment of the network.

[0013] According to a broad aspect, the present invention provides an artificial neuron network comprising a plurality of two or more layers of artificial neuron units, the layers of the artificial neuron units being configured for communication between the layers via an arrangement of two or more optical waveguides (e.g. optical fibers), the arrangement of the two or more optical waveguides being configured to have a predetermined coupling between the two or more waveguides, thereby providing cross-communication between the neuron units in the two or more layers.

[0014] In this regard, it should be noted that the term "waveguide" as used herein relates to one-dimensional waveguides, as well as two-dimensional waveguides and three-dimensional waveguides. In this regard, a one-dimensional waveguide is generally a planar waveguide that supports a single transverse mode (e.g., a single-mode optical fiber of a thin multimode optical fiber). A two-dimensional waveguide or a three-dimensional waveguide may support additional transverse modes in one or both transverse axes and, at most, may be a bulk waveguide.

[0015] According to some embodiments, at least one of the two or more optical waveguides may be configured with one or more etching patterns thereon to form one or more grating patterns to selectively enhance coupling of optical signals between the at least one waveguide and at least one other waveguide located in a selected vicinity of the etching patterns.

[0016] According to some embodiments, at least two of the two or more optical waveguides may be configured with a tapered region providing increased coupling between the at least two optical waveguides at the region.

[0017] According to some embodiments, the tapered region may further include a dedicated interaction region that provides free space interaction between optical signals propagating through the respective at least two optical waveguides associated with the tapered region. The dedicated interaction region may be formed by a ferrule element. The ferrule element may include a gain medium material that enables external pumping, thereby enabling modulation of the power of the optical signal transmitted in the ferrule element. Additionally or alternatively, the ferrule element may further include a light reflecting element at one end thereof, thereby providing backscattered light transmitted through at least one of the at least two optical waveguides associated with the tapered region.

[0018] According to some embodiments, at least two optical waveguides associated with the tapered region may further include a circulator unit that selectively defines at least one input waveguide and at least one output waveguide of the tapered region.

[0019] Typically, artificial neural networks can be configured as cascaded logic gates. In this configuration, a neural network, if formed by a topology and arrangement of neuronal units, provides a series of logic gate processing actions on a set of inputs to provide a selected output.

[0020] According to some embodiments, an artificial neuron network comprises one or more artificial neuron units, wherein at least one of the one or more artificial neuron units comprises a mode mixing unit configured to receive input light of a first wavelength range and apply selected mixing to light components of two or more spatial modes of the input light, the mode mixing unit comprising a multimode optical fiber, at least a portion of which is impregnated with a gain medium and comprises a predetermined gain medium configured to emit light of a predetermined first wavelength range in response to pump light of a second wavelength range, wherein the mode mixing unit is further configured to selectively pump additional energy to one or more spatial modes of the input light propagating through the mode mixing unit in response to the pump light of the second wavelength range and one or more selected spatial modes.

[0021] According to some embodiments, an artificial neural network includes one or more optical processing units, which include an optical gain unit having an input face and an output face located in an optical path between an input multi-core optical fiber and an output multi-core optical fiber; the optical gain unit is exposed to external illumination to generate a holographic pattern within the optical gain unit, thereby selectively affecting light transmission between the input multi-core optical fiber and the output multi-core optical fiber.

[0022] According to some embodiments, an artificial neural network includes one or more optical processing units, such optical processing units including at least one optical input port for receiving a first optical signal, at least one additional input port for receiving a second additional input signal, the optical processing unit including a first optical fiber segment, an interaction node, and a second optical fiber segment, wherein the first optical fiber segment and the second optical fiber segment are composed of optical fibers with selected properties and lengths for separating the optical signal passing therethrough into wavelength or spatial frequency components, the interaction mode being configured for receiving signal components of the first optical signal from the first optical fiber segment, interacting the signal components with the second additional input signal, and directly for generating multiplied signal components, and for coupling the multiplied signal components to the second optical fiber segment to transform the multiplied signal components and provide an output signal indicating the interaction between the first input signal and the second input signal.

[0023] According to some embodiments, the artificial neural network includes one or more processing nodes, which include an input port suitable for receiving a first input optical signal and a second input optical signal and an optical spatial mixing device, which is configured to receive the first input optical signal and the second input optical signal and apply optical processing to provide output data indicating a correlation between the first input optical signal and the second input optical signal.

[0024] According to another broad aspect, the present invention provides an artificial neuron unit, which includes a mode mixing unit, which is configured to receive input light of a first wavelength range and apply selected mixing to light components of two or more spatial modes of the input light, the mode mixing unit comprising a multimode optical fiber, at least a portion of which is impregnated with a gain medium and includes a predetermined gain medium, which is configured to emit light of a predetermined first wavelength range in response to pump light of a second wavelength range, wherein the mode mixing unit is also configured to selectively pump additional energy to one or more spatial modes of the input light propagating through the mode mixing unit in response to the pump light of the second wavelength range and one or more selected spatial modes.

[0025] According to some embodiments, the artificial neuron unit may further include a beam combiner at an input end of the artificial neuron unit, the beam combiner being positioned to guide input light of the first wavelength range and pump light of the second wavelength range to be coupled into the multimode optical fiber.

[0026] According to another broad aspect, the present invention provides a processing node (or artificial neuron unit) for an artificial neuron network, the processing node comprising an input port suitable for receiving a first input optical signal and a second input optical signal and an optical spatial mixing device configured to receive the first input optical signal and the second input optical signal and apply optical processing to provide output data indicating a correlation between the first input optical signal and the second input optical signal.

[0027] According to some embodiments, the first input optical signal and the second input optical signal may be associated with optical signals propagating in corresponding first and second multi-core optical fibers.

[0028] According to some embodiments, the optical spatial mixing device may include at least one optical reflecting element, which is configured to reflect light of the first input optical signal and the second input optical signal into a common spatial path, thereby enabling optical measurement of the spatial correlation between the first input optical signal and the second input optical signal.

[0029] According to some embodiments, the processing node may further comprise a decoherence unit configured to reduce the spatial coherence of the first input optical signal and the second input optical signal upstream of the optical spatial mixing device.

[0030] According to some embodiments, the processing node can be configured to receive the first input optical signal and the second input optical signal through the first multi-core optical fiber and the second multi-core optical fiber, and the optical spatial mixing device is configured to mix the first input optical signal and the second input optical signal in free space propagation of light.

[0031] According to another broad aspect, the present invention provides an optical processing unit (or artificial neuron unit), which includes an optical gain unit having an input face and an output face located in an optical path between an input multi-core optical fiber and an output multi-core optical fiber; the optical gain unit is exposed to external illumination to generate a holographic pattern within the optical gain unit, thereby selectively affecting the light transmission between the input multi-core optical fiber and the output multi-core optical fiber.

[0032] According to some embodiments, the holographic pattern within the optical gain unit may be three-dimensional.

[0033] According to some embodiments, the optical processing unit may further include an input optical lens unit and an output optical lens unit, wherein the input optical lens unit and the output optical lens unit are respectively located in the optical path between the input multi-core optical fiber and the input surface of the optical gain unit and between the output surface of the optical gain unit and the output multi-core optical fiber.

[0034] According to some embodiments, the input multi-core fiber and the output multi-core fiber may be formed as one-dimensional multi-core fibers, and the input optical lens unit is an astigmatic lens configured to guide input light to form a three-dimensional spatial pattern within the optical gain unit.

[0035] According to another broad aspect, the present invention provides an optical processing unit comprising at least one optical input port for receiving a first optical signal, at least one additional input port for receiving a second additional input signal, the optical processing unit comprising a first optical fiber segment, an interaction node and a second optical fiber segment, wherein the first optical fiber segment and the second optical fiber segment are composed of optical fibers with selected properties and lengths for separating the optical signal passing therethrough into wavelength or spatial frequency components, the interaction mode being configured for receiving a signal component of the first optical signal from the first optical fiber segment, interacting the signal component with the second additional input signal and directly for generating a multiplied signal component, and for coupling the multiplied signal component to the second optical fiber segment to transform the multiplied signal component and provide an output signal indicating the interaction between the first input signal and the second input signal.

[0036] According to some embodiments, the first and second fiber segments may include graded-index fiber segments whose refractive index profile and length are selected to separate optical signals passing through the graded-index fiber segments into spatial frequencies of the optical signals, thereby applying a spatial Fourier transform to the input signals.

[0037] According to some embodiments, the interaction mode may include an arrangement of multiple optical fiber cores formed by optical fibers carrying gain material, and wherein the second additional input signal provides selective pumping to the arrangement of multiple optical fiber cores, thereby selectively causing components of the second additional input signal to interact with spatial components of the first optical signal.

[0038] According to some embodiments, the first optical fiber segment and the second optical fiber segment may include a dispersive optical fiber having a length selected to apply a Fourier transform to the optical signal with respect to wavelength components of the optical signal, and the interaction node includes a time modulator configured to receive data indicative of the second additional input signal and modulate components of the first optical signal accordingly, thereby causing frequency components of the first input signal and the second input signal to interact.

[0039] This application provides the following: 1) An artificial neuron network comprising a plurality of two or more layers of artificial neuron units, the layers of artificial neuron units being configured to communicate between the layers via an arrangement of two or more optical waveguides, the arrangement of the two or more optical waveguides being configured to have a predetermined coupling between the two or more waveguides, thereby providing cross communication between the neuron units of the two or more layers.

[0040] 2) The artificial neural network according to 1), wherein at least one of the two or more optical waveguides is provided with one or more etching patterns, and the one or more etching patterns form one or more grating patterns, thereby selectively enhancing the coupling of optical signals between the at least one waveguide and at least one other waveguide located in a selected vicinity of the etching patterns.

[0041] 3) The artificial neural network according to 1) or 2), wherein at least two of the two or more optical waveguides are configured with a tapered region at which increased coupling is provided between the at least two optical waveguides.

[0042] 4) The artificial neural network according to 3), wherein the tapered region further comprises a dedicated interaction region providing free-space interaction between optical signals propagating through the respective at least two optical waveguides associated with the tapered region.

[0043] 5) The artificial neural network according to 4), wherein the dedicated interaction area is formed by a ferrule element.

[0044] 6) The artificial neural network according to 5), wherein the ferrule element comprises a gain medium material, and the gain medium material can be externally pumped to modulate the power of the optical signal transmitted in the ferrule element.

[0045] 7) The artificial neural network according to 5) or 6), wherein the ferrule element further comprises a light reflecting element at one end of the ferrule element, thereby providing backscattered light transmitted through at least one of the at least two optical waveguides associated with the tapered region.

[0046] 8) The artificial neural network according to 7), wherein the at least two optical waveguides associated with the tapered region further comprise a circulator unit, the circulator unit selectively defining at least one input waveguide and at least one output waveguide of the tapered region.

[0047] 9) An artificial neuron network according to any one of 1) to 8), comprising one or more artificial neuron units, wherein at least one of the one or more artificial neuron units comprises a mode mixing unit, the mode mixing unit being configured to receive input light of a first wavelength range and apply a selected mixing to optical components of two or more spatial modes of the input light, the mode mixing unit comprising a multimode optical fiber, at least a portion of the multimode optical fiber being impregnated with a gain medium and comprising a predetermined gain medium, the predetermined gain medium being configured to emit light of a predetermined first wavelength range in response to pump light of a second wavelength range, wherein the mode mixing unit is further configured to selectively pump additional energy into one or more spatial modes of the input light propagating through the mode mixing unit in response to the pump light of the second wavelength range and one or more selected spatial modes.

[0048] 10) An artificial neural network according to any one of 1) to 9), comprising at least one optical processing unit, wherein the optical processing unit includes an optical gain unit, wherein the optical gain unit has an input surface and an output surface located in an optical path between an input multi-core optical fiber and an output multi-core optical fiber; the optical gain unit is exposed to external illumination to generate a holographic pattern within the optical gain unit, thereby selectively affecting light transmission between the input multi-core optical fiber and the output multi-core optical fiber.

[0049] 11) An artificial neural network according to any one of 1) to 10), comprising at least one optical processing unit, the optical processing unit comprising at least one optical input port for receiving a first optical signal, at least one additional input port for receiving a second additional input signal, the optical processing unit comprising a first optical fiber segment, an interaction node and a second optical fiber segment, wherein the first optical fiber segment and the second optical fiber segment are composed of optical fibers with selected characteristics and lengths for separating the optical signals passing through the first optical fiber segment and the second optical fiber segment into wavelength or spatial frequency components, the interaction mode is configured for receiving a signal component of the first optical signal from the first optical fiber segment, causing the signal component to interact with the second additional input signal and directly for generating a multiplied signal component, and for coupling the multiplied signal component to the second optical fiber segment for transforming the multiplied signal component and providing an output signal indicating the interaction between the first input signal and the second input signal.

[0050] 12) An artificial neural network according to any one of 1) to 11), comprising at least one processing node, the processing node comprising an input port suitable for receiving a first input optical signal and a second input optical signal and an optical spatial mixing device, the optical spatial mixing device being configured to receive the first input optical signal and the second input optical signal and apply optical processing to provide output data indicating a correlation between the first input optical signal and the second input optical signal.

[0051] 13) The artificial neural network according to any one of 1) to 12), configured to operate as cascaded logic gates.

[0052] 14) An artificial neuron unit, comprising a mode mixing unit, the mode mixing unit being configured to receive input light of a first wavelength range and to apply selected mixing to optical components of two or more spatial modes of the input light, the mode mixing unit comprising a multimode optical fiber, at least a portion of which is impregnated with a gain medium and comprises a predetermined gain medium, the predetermined gain medium being configured to emit light of a predetermined first wavelength range in response to pump light of a second wavelength range, wherein the mode mixing unit is further configured to selectively pump additional energy into one or more spatial modes of the input light propagating through the mode mixing unit in response to the pump light of the second wavelength range and one or more selected spatial modes.

[0053] 15) The artificial neuron unit according to 14), further comprising a beam combiner located at an input end of the artificial neuron unit, the beam combiner being positioned to guide the input light of the first wavelength range and the pump light of the second wavelength range to be coupled into the multimode optical fiber.

[0054] 16) A processing node for use in an artificial neural network, the processing node comprising an input port adapted to receive a first input optical signal and a second input optical signal and an optical spatial mixing device configured to receive the first input optical signal and the second input optical signal and to apply optical processing to provide output data indicative of a correlation between the first input optical signal and the second input optical signal.

[0055] 17) The processing node according to 16), wherein the first input optical signal and the second input optical signal are associated with optical signals propagating in corresponding first and second multi-core optical fibers.

[0056] 18) A processing structure according to 16) or 17), wherein the optical spatial mixing device includes at least one optical reflecting element, and the optical reflecting element is configured to reflect light of the first input optical signal and the second input optical signal into a common spatial path, thereby enabling optical measurement of the spatial correlation between the first input optical signal and the second input optical signal.

[0057] 19) The processing node according to any one of 16) to 18), further comprising a decoherence unit configured to reduce the spatial coherence of the first input optical signal and the second input optical signal upstream of the optical spatial mixing device.

[0058] 20) A processing junction according to any one of 16) to 19), wherein the processing junction is configured to receive the first input optical signal and the second input optical signal through a first multi-core optical fiber and a second multi-core optical fiber, and the optical spatial mixing device is configured to mix the first input optical signal and the second input optical signal in free space propagation of light.

[0059] 21) An optical processing unit, comprising an optical gain unit, the optical gain unit having an input face and an output face located in an optical path between an input multi-core optical fiber and an output multi-core optical fiber; the optical gain unit is exposed to external illumination to generate a holographic pattern within the optical gain unit, thereby selectively affecting light transmission between the input multi-core optical fiber and the output multi-core optical fiber.

[0060] 22) The optical processing unit according to 21), wherein the holographic pattern within the optical gain unit is three-dimensional.

[0061] 23) The optical processing unit according to 21) further includes an input optical lens unit and an output optical lens unit, wherein the input optical lens unit and the output optical lens unit are respectively located in the optical path between the input multi-core optical fiber and the input surface of the optical gain unit and in the optical path between the output surface of the optical gain unit and the output multi-core optical fiber.

[0062] 24) The optical processing unit according to 23), wherein the input multi-core optical fiber and the output multi-core optical fiber are formed as one-dimensional multi-core optical fibers, and the input optical lens unit is an astigmatic lens configured to guide input light to form a three-dimensional spatial pattern within the optical gain unit.

[0063] 25) An optical processing unit comprising at least one optical input port for receiving a first optical signal, at least one additional input port for receiving a second additional input signal, the optical processing unit comprising a first optical fiber segment, an interaction node and a second optical fiber segment, wherein the first optical fiber segment and the second optical fiber segment are composed of optical fibers with selected characteristics and lengths for separating the optical signals passing through the first optical fiber segment and the second optical fiber segment into wavelength or spatial frequency components, the interaction mode is configured for receiving a signal component of the first optical signal from the first optical fiber segment, causing the signal component to interact with the second additional input signal and directly for generating a multiplied signal component, and for coupling the multiplied signal component to the second optical fiber segment for transforming the multiplied signal component and providing an output signal indicating the interaction between the first input signal and the second input signal.

[0064] 26) An optical processing unit according to 25), wherein the first optical fiber segment and the second optical fiber segment include a graded-index optical fiber segment having a selected refractive index distribution and length for separating an optical signal passing through the graded-index optical fiber segment into spatial frequencies of the optical signal, thereby applying a spatial Fourier transform to the input signal.

[0065] 27) An optical processing unit according to 26), wherein the interaction mode includes an arrangement of multiple optical fiber cores formed by optical fibers carrying gain material, and wherein the second additional input signal provides selective pumping to the arrangement of multiple optical fiber cores, thereby selectively causing components of the second additional input signal to interact with spatial components of the first optical signal.

[0066] 28) An optical processing unit according to any one of 25) to 27), wherein the first optical fiber segment and the second optical fiber segment include dispersive optical fibers having a selected length to apply a Fourier transform to the optical signal with respect to a wavelength component of the optical signal, and the interaction node includes a time modulator configured to receive data indicative of the second additional input signal and to modulate a component of the first optical signal accordingly, thereby causing the frequency components of the first input signal and the second input signal to interact. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to better understand the subject matter disclosed herein and to illustrate how it may be carried out in practice, embodiments will now be described, by way of non-limiting examples only, with reference to the accompanying drawings, in which: Figure 1 The overall structure of artificial neural networks is illustrated; Figure 2illustrates coupling optical signals between optical fibers according to some embodiments of the present invention; Figure 3 An artificial neuron unit capable of nonlinear processing according to some embodiments of the present invention is shown; Figure 4 An artificial neuron unit configured to determine correlation between input signals according to some embodiments of the present invention is shown; Figure 5A and Figure 5B shows an optical convolution and transformation unit according to some embodiments of the present invention, Figure 5A illustrates a spatial convolution unit, and Figure 5B 3D spatial and temporal convolution units are illustrated; FIG. 6A to FIG. 6C An example of an artificial neuron unit configuration using a bulk gain structure according to some embodiments of the present invention is shown. Fig. 6A 3D patterning of bulk gain structures and direct coupling is illustrated, Figure 6B demonstrates free-space coupling that allows two-dimensional patterning of bulk gain structures, and Figure 6C The use of bulk gain structures and patterns thereof as optical switching elements is illustrated; and Fig. 7A and Figure 7B An artificial neuron unit utilizing a tapered multi-core optical fiber for mixing selected signal portions and applying gain thereto is shown according to some embodiments of the present invention. DETAILED DESCRIPTION

[0068] Artificial neural networks refer to a variety of computing architectures and algorithms inspired by biological neural networks. Figure 1 , Figure 1 Illustrated is a schematic configuration of an artificial neuron network 100 formed by an arrangement of artificial neuron units (neurons) 200 , which communicate via links 210 between them.

[0069] In general, the neurons 200 are arranged in layers including an input layer INL positioned and configured to receive input data for processing and an output layer OL positioned and configured to provide output data after processing. The network may also include one or more intermediate hidden layers, such as Figure 1 HL1 and HL2 in the example. One or more intermediate layers of neurons bridge between the input layer and the output layer and participate in data processing.

[0070] As described above, optical processing and optical neural networks can provide enhanced processing speeds compared to electronic processing. Typically, implementations of optical (photonic) neural networks are based on two primary operations, such as linear mixing between signals associated with two or more artificial neurons (i.e., propagating along two or more optical fibers or cores) and nonlinear effects, such as those associated with amplification and mixing, to provide processing within the artificial neuron unit.

[0071] Artificial neural networks can be configured and trained to perform various computing operations. Typically, artificial neural networks such as network 100 can be configured for specific processing applications based on the topology of the network, the number of layers, the arrangement of the layers, and network training for specific applications. For example, artificial neural network 100 can be configured to operate as a cascade logic gate. Such logic gates can determine output data based on a plurality of different inputs and the selected relationship between them. Additional exemplary applications may include various data processing directly applied to optical inputs, including processing applications such as image recognition. It should be noted that the present invention relates to the configuration of optical neural networks and artificial neuron units, and can be used for any processing application to which the network and neuron configuration of the present invention are suitable.

[0072] According to some embodiments, the present technology provides an artificial optical neuron unit suitable for operating in an artificial optical neuron network. The artificial neuron unit is configured based on one or more optical fibers or waveguides, and is configured to provide amplification gain, mixing and / or applying selected operations / transformations to optical signals. The present technology eliminates or at least significantly reduces the need for photoelectric conversion of signals, and thus generally avoids the need for photoelectric signal conversion. Artificial neuron units according to the present technology are typically formed using one or more optical fibers, which are configured and used for at least one of the following items: transmitting optical signal portions, coupling optical signal portions between artificial neuron units, and applying one or more processing operations to optical signal portions, thereby operating one or more artificial neuron units.

[0073] refer to Figure 2, which illustrates an optical link unit 210a formed by an arrangement of optical fibers 220a to 220n. The optical link unit provides transmission of an optical signal to provide an input signal to an artificial neuron unit. The optical link unit 210a is formed in one or more selected optical fiber sections near one or more other optical fiber sections. In this particular example, the optical fiber 220b is etched with a periodic pattern 250, forming a grating region (e.g., a Bragg grating) with a selected period and length (e.g., a long period Bragg grating). The periodic pattern 250 is selected to couple out a pre-selected signal portion 260 from the optical fiber 220b. The signal portion 260 coupled out from the optical fiber 220b can be coupled to an adjacent optical fiber, such as 220a and / or 220c, for mixing the signal portion. Additionally or alternatively, the signal portion 260 or a portion thereof can be directed to the outside of the optical fiber arrangement, forming a loss of a selected portion of the optical signal. By selecting parameters of the periodic pattern 250, the signal portion 260 can be selected based on wavelength and / or spatial mode.

[0074] The periodic pattern 250 may be configured as a long period fiber Bragg grating (LPFBG) applied (e.g., etched) on a portion of one or more selected optical fibers 220b or on a portion of the optical fiber core (in the case of a multi-core optical fiber bundle). The pattern 250 provides controlled coupling between optical fibers using weights for mixing input signal portions selected by the parameters of the pattern 250. The use of the periodic pattern 250 allows controlled coupling between optical fibers while reducing the requirements of the geometric relationship, such as the length and physical distance between adjacent optical fibers.

[0075] In some additional configurations, the optical link unit 210a may utilize a periodic pattern 250 formed as a short period fiber Bragg grating (SPFBG). The period of the SPFBG is selected according to the wavelength of the optical signal so as to separate the signal portion based on the wavelength of the signal passing through the optical fiber 220b. The use of the SPFBG along the selected optical fiber allows the use of multi-spectral optical signals and selectively separates the signal portion propagating through the optical fiber based on its wavelength. Such a periodic pattern 250 can be advantageously used with an optical fiber doped with a gain material for selectively amplifying the signal portion, such as an erbium doped fiber (EDFA), which will be described in more detail below. For example, in such a gain-doped fiber, the pump wavelength can be 980nm and the signal wavelength can be approximately 1550nm, and the use of the SPFBG allows the pump wavelength and the signal wavelength to be separated by back-reflecting the pump wavelength while allowing the signal wavelength to propagate without change.

[0076] In some configurations, the core of one or more optical fibers (220b) may be provided with a selected active material having nonlinear properties. Such active material provides an optical effect that selectively changes the periodic pattern, for example by changing the refractive index profile along the fiber. For example, the core of one or more optical fibers 220b may be formed of or doped with a photorefractive material (i.e., a material that exhibits a photorefractive effect), a material having Kerr nonlinearity, or a liquid crystal material that allows the refractive index of the material to be selectively changed. The refractive index may change in response to optical interactions caused by nonlinear interactions in the active material or using an external field applied to the optical fiber 220b. For example, the use of such an active material allows the optical properties of the periodic pattern 260 to be selectively changed by external illumination or by applying an external electric field thereto, and in this case, the recording of the LPFBG via external illumination may be reconfigurable or tunable, for example by changing the refractive index and thus affecting the wavelength of the signal compared to the periodic pattern 250.

[0077] As shown, the artificial neuron unit 200 can be configured based on a multimode optical fiber (MMF). This multimode optical fiber (MMF) is configured to have a first end and a second end and a selected length and diameter, and is used to propagate an input signal therethrough while mixing the spatial modes of the propagating signal. More specifically, the light field input to the MMF can be a combination of one or more spatial modes with respect to the MMF structure. Each spatial mode propagates through the MMF at a corresponding group velocity, changing the modal combination of the output light. In addition, propagation through the MMF can result in some mixing between light components, thereby transferring light energy between modes according to the shape and optical properties of the MMF. Therefore, the MMF provides an outgoing signal associated with a mode mixing of the input optical signal.

[0078] Furthermore, according to some embodiments, the MMF may be doped with a gain material (eg, erbium-doped MMF) selected to provide gain in a wavelength range associated with the wavelength of the optical signal passing therethrough. In this regard, Figure 3An artificial neuron unit 200 according to some embodiments of the present invention is schematically shown. The artificial neuron unit 200 is based on a multimode optical fiber (MMF) 10 doped with a selected amount of a gain medium, which is configured to emit light of a first wavelength range in response to pump light of a second (pump) wavelength range. The MMF 10 has a first end for receiving input light and a second end for providing output light OL. The MMF 10 can generally be configured as an optical fiber with a single wide core that supports multiple spatial optical modes of the first wavelength range. The artificial neuron unit 200 may also include a beam splitter unit BS configured to receive a first input light wavefront WF associated with an optical signal and a second pump light PL. The input light wavefront WF may be formed by a specific spatial light field. The pump light PL may have a selected spatial distribution to provide pump energy to one or more selected spatial modes of light within the MMF 10. The artificial neuron unit 200 may also include an input optical device 20 and an output optical device 30, the input optical device 20 being configured to couple light into the MMF 10. Furthermore, the artificial neuron unit 100 may include a spatial light modulator 40 located in the optical path of the light coupled out from the MMF 10, which may also be associated with a control unit configured to selectively change the modulation pattern of the spatial light modulator 40. It should be noted that the input light and the output light of the artificial neuron unit 200 may come from free space propagation light and / or light propagation between linked optical fibers.

[0079] The artificial neuron unit 200 is configured to receive an input light WF signal, typically coupled into the MMF 10 by an input optical device 20, propagate the input light WF through the MMF 10, mix the spatial modes of the input light signal WF to a certain degree, and provide an output light EL at its second end. In addition, when a selected spatial distribution of pump light PL is input to the MMF 10, the pump light is used to selectively amplify the signal portion associated with the spatial mode having a high degree of overlap with the spatial distribution of the pump light PL. The spatial light modulator 40 may also selectively modulate the output light EL according to the selected operation / task of the artificial neuron unit, for which the neuron unit is trained, providing an output light signal OL. In this regard, it should be noted that processing techniques typically using a neural type configuration are based on one or more networks of neuron units. The network preferably undergoes a selected training process, in which internal connections, processing parameters are determined. The artificial neuron unit 100 described herein can be used in various network topologies. For simplicity, the artificial neuron unit 100 may be configured as a single processing unit, where a selected optical operation may be performed by mixing the spatial modes of the input optical signal WF and amplifying the selected spatial mode using the pump light PL.

[0080] The MMF 10 is a multimode optical fiber having a selected length (e.g., several millimeters to several centimeters, in some embodiments, the MMF 10 can be up to several meters) and diameter (e.g., 30 microns or more, 50 microns or more), and is generally configured to support the propagation of light in a selected wavelength range (e.g., 1.5 microns) propagating in multiple spatial modes. The core of the MMF 10 is doped with a material having gain characteristics at a selected doping ratio, such as providing the MMF 10 with erbium or other rare earth doping.

[0081] Typically, an input optical signal having specific wavefront WF, amplitude and length characteristics is transmitted to the artificial neuron unit 100. The input optical signal WF is coupled into the MMF 10 through the input optical device 20 and propagates in the MMF 10 toward its second end. In addition, pump light PL of a selected spatial waveform (mode) is also coupled into the MMF 10 to provide optical pumping of the gain medium embedded in the MMF 10. While propagating through the MMF, the mode with high spatial correlation of the pump light 10 is amplified by the stimulated emission of the gain medium. In addition, optical signals of different spatial modes (corresponding to the spatial shape of the input light wavefront WF projected onto the structure of the MMF 10) propagate at different speeds and experience mixing between them. The length of the MMF 10 is selected according to the desired mixing between the spatial modes and the amplification level provided thereby. Typically, for a relatively short MMF 10, ie short relative to the group velocity dispersion characteristics of the MMF 10, the outgoing light EL retains most of the mode content of the input wavefront WF, wherein selected modes are amplified according to the spatial wavefront of the pump light PL.

[0082] The exit light EL may be directed to a spatial light modulator 40 which applies a selected spatial modulation to the wavefront, thereby providing an output light OL signal. The output light OL signal may then be directed to one or more additional neuronal units 200 associated with additional layers of the network, and / or to corresponding detection units.

[0083] The input optical device 20 may be located near the input end of the MMF 10 and configured to couple the input light WF into the MMF 10. In some configurations, the input optical device may be configured to also couple the pump light into the MMF 10, which may be achieved by placing the optical device 20 downstream of the beam splitter unit BS with respect to the propagation of the light into the MMF 10. Optionally, additional optical devices (not specifically shown) may be used to couple the pump light into the MMF 10. In general, the input optical device 20 may include one or more optical elements, such as one or more lenses (e.g., an objective lens unit).

[0084] As described above, in some configurations, the artificial neuron unit 200 may further include an output optical device 30 located downstream of the MMF 10, such as between the MMF 10 and the spatial light modulator (SLM) 40 and / or downstream of the SLM 40, such as Figure 3 As shown. The output optical device 30 can generally be composed of one or more optical elements (such as lenses). The output optical device is generally configured to collect the output light OL from the artificial neuron unit, and influence the divergence and / or propagation direction of the output light (for example, to provide collimated output light) according to the selected path of the output light toward the additional neuron, and couple to another optical fiber and / or one or more detection units.

[0085] For example, in a typical communication system, the optical signal used has a first wavelength range of about 1550 nm. In addition, a typical erbium-doped fiber can respond to pump light of a second wavelength of about 980 nm by emitting light of about 1550 nm corresponding to the first wavelength range. Therefore, appropriate shaping of the pump light PL spatial distribution can provide selective pumping of one or more selected spatial modes of the optical signal traveling through the MMF 10.

[0086] It should be noted that the artificial neuron unit 200 described herein may also be patterned with one or more periodic patterns selected to couple out the residual pump light or cause back reflection of the residual pump light to reduce interference. Such periodic patterns are described above with reference to Figure 2 Alternatively or additionally, the artificial neuron unit 200 or any optical link unit collecting the output light OL may include a spectral filter selected to allow transmission of wavelengths of a first wavelength range while filtering out pump light of a second wavelength range. The spectral filter may be configured to absorb or deflect the pump light to prevent the pump light from being transmitted between layers of the neural network.

[0087] Certain processing operations may be associated with determining the correlation between two input signals. For example, such a processing step may be part of training a neural network. Figure 4, which shows an optical processing unit 240, which is used in an artificial neural network to determine the correlation between two input signals (for example, a processed unlabeled data segment relative to a processed labeled data segment). The optical processing unit 240 can operate as an artificial neuron unit 200 within a neural network, which has selected processing operations and is configured to receive a first input optical signal ILa and a second input optical signal ILb, for example, through corresponding multi-core optical fiber bundles 12a and 12b. The optical processing unit 240 is configured to spatially fold the first optical signal ILa and the second optical signal ILb, providing an output light OL indicating spatial interference between the signals. The output signal OL can be detected by the optical detector 26, or directed to other processing layers for additional processing.

[0088] exist Figure 4 In a specific example of, the optical processing unit 240 includes a lens unit 20 having a selected focal length f and a beam splitter unit BS, which is positioned to receive the input light IL after passing through the lens unit 20. Typically, the first part of the correlator optical processing unit 240 has spatially coherent light and a Fourier transform performed by the lens 20. A rotating diffuser 28 can be placed in the optical path of the input light to destroy the phase and convert the spatially coherent light into a spatially incoherent signal distribution. As illustrated, the rotating diffuser 28 can be placed between the lens unit 20 and the beam splitter BS. The second part of the correlator optical processing unit 240 is a cosine transform, which is performed using a shearing interferometer formed by the beam splitter BS, the corner prism 22 and the reflective surface (mirror) 24. The output light OL has the form of a correlation peak in terms of the spatial correlation between the two input light signals ILa and ILb. The output light OL can be detected by a single pixel detector 26, or transmitted for further use.

[0089] The lens unit 20 is preferably positioned to provide a Fourier image of the first input light signal ILa and the second input light signal ILb onto a rotating diffuser 28 (i.e., with a distance f between the outputs of the optical fibers 12a and 12b). The light distribution IL may become incoherent due to the rotating diffuser 28 and is directed to the beam splitter unit BS. The beam splitter unit BS reflects at least a portion R1 of the input light IL towards a corner prism 22, which is positioned to fold the light pattern around the optical axis of the prism 22, directing the folded light portion R2. The light pattern thus folded is directed R3 to a reflecting surface (mirror) 24 and is reflected again R4 to the beam splitter BS providing an output light OL, wherein the first portion ILa and the second portion ILb of the input light IL are folded against each other and interfere with each other.

[0090] Typically, when the first input signal ILa and the second input signal ILb match, the output light OL provides a high correlation peak output. If the input signals are different, the output light OL provides a low reading. Figure 4 The optical processing unit 240 illustrated in the figure can be used, for example, in the training process of an optical neural network to determine correlations between optical signals undergoing processing, while eliminating or at least significantly reducing the need to convert the optical signals into electronic data.

[0091] Some processing functions may be associated with determining the frequency content of a signal, or generally with determining the Fourier transform of a signal. This may be associated with separate processing of the temporal or spatial frequency components of a signal and performing certain operations such as convolution of two signals. To this end, an artificial neural network may utilize one or more artificial neural units, including a selected length of graded-index optical fiber (GRIN fiber) having a selected refractive index profile and length for affecting a light pattern propagating through the fiber segment to perform one or more selected transformations on the light pattern. For example, the refractive index profile and length may be selected for performing an optical Fourier transform on at least one of the spatial and temporal dimensions of an optical signal coupled thereto. Performing a selected spatial or temporal transform on the input optical signal may be used for additional operations that can be performed directly on the optical signal. For example, referring to Figure 5A and Figure 5B , which illustrates performing spatial convolution ( Figure 5A ) and spatial and temporal convolution ( Figure 5B ) is an optical convolution unit 200. In this example, the optical convolution unit 200 utilizes a spatial Fourier transform and an erbium doped fiber (EDFA) to determine the spatial convolution between the input optical signal IL and the selected weights of the spatial frequencies FW provided by the selected gain stage (e.g., by the selected pump). The product of the signals is transformed into the convolution of the Fourier transformed signals, and vice versa. The optical convolution unit 200 is formed by a multi-mode graded refractive index (GRIN) fiber segment 50a, followed by an EDFA network 512, which is configured to receive a pump level with a selected radial distribution FW, which is associated with selected weights applied to different spatial frequencies (e.g., determined by the pump intensity). The EDFA network 512 may be followed by an additional GRIN fiber segment 50b to provide an output optical signal OL.

[0092] In some examples, the GRIN fiber segments 50a and 50b may preferably have a refractive index profile of the following form: (Equation 1) Where r is a radial coordinate within the fiber, and n1 and n2 are selected refractive index values ​​of the fiber. Using this refractive index profile, fiber segments 50a and 50b can be configured to have lengths When an optical signal propagates through the optical fiber segment 50a or 50b, changes in the refractive index and phase velocity of the signal portion result in a two-dimensional spatial Fourier transform of the signal waveform.

[0093] Thus, an input signal IL having an input waveform structure can be coupled into the first GRIN fiber segment 50a. The input signal propagates through the GRIN fiber segment 50a, wherein the signal portion propagates at a corresponding group velocity, so that at the output end of the GRIN fiber segment 50a, the intermediately generated signal is the spatial Fourier transform of the input signal. The intermediately generated signal is coupled into the EDFA network 512, so that different spatial frequencies are coupled into different radial portions of the network 512. As described above, the EDFA network 512 is configured to provide a selected radial gain profile (e.g., using a gradient doping level or a selected pumping profile), thereby applying a selected gain level to different Fourier components of the input signal, providing a multiplied Fourier signal. The multiplied Fourier signal is further coupled into the GRIN fiber segment 50b and propagates therein while undergoing an inverse Fourier transform to provide an output light OL. The output light is a convolution form between the input optical signal and the inverse Fourier transform of the gain profile in the EDFA network 512.

[0094] Typically, the EDFA network 512 may include an arrangement of NxN fibers appropriately connected to the output end of the GRIN fiber segment 50a and the input end of the GRIN fiber segment 50b to couple optical segments between the fiber segments. There are NxN points in the output plane of the input GRIN fiber and NxN points in the input plane of the output GRIN fiber. The selected number and arrangement of the NxN fibers of the EDFA network 512 is related to the resolution of the Fourier transform and convolution processes.

[0095] A dispersive fiber provides varying phase velocities for light of different wavelengths passing through it. Such a dispersive fiber can therefore apply a selected transformation to an input optical signal with respect to its temporal frequency. Typically, such a dispersive fiber performs a temporal Fresnel transform on an input optical signal coupled to it. A sufficiently long dispersive fiber (i.e., one long enough to satisfy the far-field approximation) will affect the optical signal passing through it as a Fourier transform in the time domain. Figure 5BAn optical convolution unit 200 is shown, which is configured to provide temporal convolution and spatial convolution of a signal. In this regard, the intermediately generated signal output from the GRIN fiber segment 50a is coupled to a temporal convolution portion formed by a long dispersion fiber 60a (length not shown to scale), a time modulator 612, and a second long dispersion fiber 60b. The temporal convolution portion provides for determining the convolution between the input signal (intermediately generated signal) and the selected modulated Fourier applied by the time modulator 612. More specifically, the long dispersion fibers 60a and 60b affect the input signal to provide its Fourier transform (or inverse Fourier transform). The time modulator 612 multiplies the temporal frequency with the selected modulation time pattern, which, after the inverse Fourier transform, provides a convolution of the signal. The temporal convolution is between the temporal variation of the input signal and the inverse Fourier of the signal fed to the time modulator 612.

[0096] It should be noted that such GRIN fiber segments (e.g., 50a) as well as long dispersion fibers (e.g., 60a) can be used to form additional processing and transformation units in addition to the convolution units illustrated herein. More specifically, using Figure 5A or Figure 5B Half of the arrangement, i.e. removing the time modulators 512 and 612 and replacing them with output terminals, enables the spatial and / or temporal Fourier transform of the selected signal to be determined. Such a transformed signal can also be used for additional processing actions applied to its selected frequency components. In addition, additional processing elements (e.g. in Figure 3 The selective modal gain processing 200 illustrated in FIG. 2 ) can be used as an intermediate component of such a transform and convolution unit 200 .

[0097] For example, in some configurations, the convolution unit 200 can be configured to generate a modulated clock signal output OL. More specifically, the input clock signal can be used for the input signal IL, wherein the EDFA network 512 can apply spatial modulation to the clock signal. Additionally or alternatively, the time modulator 612 can be used to apply a selected time modulation or wavelength modulation to encode the clock signal in time or in wavelength. Such a clock signal can also be obtained by further processing using the three-dimensional space-time transform / convolution unit 200. A selected wavelength selective filter can be used to filter out unwanted wavelengths from the time Fourier component for applying further processing operations to the frequency components of the input signal IL.

[0098] In some additional configurations of the present technology, the optical processing unit may include a waveguide pattern selectively written / engraved into the bulk gain structure. Typically, such a selected pump pattern may be optically applied to the bulk gain structure to provide a selected amplification / light manipulation effect depending on the pump pattern. Such a pump pattern may also provide writing of a waveguide pattern within the bulk gain structure, enabling selective filtering and phase velocity manipulation of the input optical signal. Reference FIG. 6A to FIG. 6C , illustrates an artificial neuron unit 200 arrangement that utilizes a bulk gain structure 17 subjected to a selective writing pattern HW (e.g., volume holographic writing) to provide a selected waveguide and a selected pump arrangement patterned therethrough. The artificial neuron unit 200 comprises an input multi-core fiber 70a and an output multi-core fiber 70b, which are configured to receive an input optical signal IL and guide the input optical signal to couple into the bulk gain structure 17 and to receive an optical signal leaving the bulk gain structure 17 and guide it to provide an output optical signal OL. Fig. 6A In the example of FIG. 1 , the coupling between the multicore waveguides 70a and 70b and the gain structure 17 is direct, ie, the optical fiber is coupled to the bulk waveguide. Figure 6B In the example of FIG. 1 , optical lenses 20 and 30 are used to couple light between multi-core waveguides 70a and 70b and gain structure 17. Figure 6C In the example of FIG. 1 , the body structure can be selectively used as a switch, which allows the input light to be selectively introduced into the multi-core waveguide 70 b or the multi-core waveguide 70 c to provide the first output light OL1 or the second output light OL2 signal.

[0099] The selected write pattern HW can provide two-dimensional or three-dimensional writing of waveguide channels within the gain structure 17. This can be performed using a fast laser system to provide a configurable and controllable architecture in which the waveguide channels and their gain levels can be selectively changed depending on the light input used to write the HW. Figure 6C As shown, selective writing can also be used for selective routing of optical signals. More specifically, depending on the current written HW pattern on the gain structure 17 and the signal interaction with the so-generating pattern within the gain structure 17, the input signal IL can be directed to the output signal OL1 or OL2.

[0100] Typically, the bulk gain structure 17 may be patterned using three-dimensional writing. However, in some configurations, the pattern may be two-dimensional, allowing for simpler pattern configuration. Figure 6B The example allows the following configuration: wherein the multi-core optical fiber 70a and / or 70b is a one-dimensional optical fiber array carrying a one-dimensional optical signal. In this configuration, the optical lenses 20 and 30 can be astigmatic lenses configured to guide a relatively wide one-dimensional pattern relative to the bulk gain structure 17 to avoid divergence of the optical signal.

[0101] In another exemplary configuration according to the present technology, a tapered multi-core optical fiber can be used to provide direct and appropriate interaction between the optical fiber cores. Fig. 7A and Figure 7B , which illustrate a controllable artificial neuron unit 200 coupled using a tapered multi-core optical fiber having a tapered configuration. In general, a tapered optical fiber configuration enables interaction between optical fibers, such that away from the tapered region, the fiber cores are separated to prevent interaction, while the tapered region brings the cores together to allow interaction between optical signals transmitted therethrough, such as through optical coupling. Fig. 7A and Figure 7B An artificial neuron unit 200 is shown having input and output multi-core optical fibers 80a and 80b, which are configured with Fig. 7A The tapered regions 82a and 82b ( Figure 7B 82) to provide light coupling between the cores of the multi-core optical fiber. Fig. 7A In the example of , an input optical signal having a particular input waveform propagates through an input multi-core fiber 80a, and as the multi-core fiber tapers to a ferrule 85, the light is allowed to propagate in a free space mode through the ferrule connector 85, and is then coupled to an output multi-core fiber 80b, which allows coupling and mixing between light components propagating in different cores of the multi-core fiber 80b. The use of a ferrule 85 after the tapered region 82a provides free space interaction between spatial modes and increases mixing. In addition, the ferrule may include or be made of a nonlinear material having selected optical properties, such as a gain medium, an optical Kerr effect, etc., which are selected to allow selective pumping or alteration of mixing properties of portions of the signal. For example, the ferrule 85 may include an erbium gain medium or other rare earth material to allow external pumping and control of selected weights in a neural network. In some additional configurations, erbium fiber may be used for the input multi-core fiber 80a and / or the output multi-core fiber 80b, with the ferrule 85 providing free space mixing between the optical components such that the input optical signal is first amplified and then mixed and interacted via the tapered device 82a and via the ferrule 85.

[0102] in this regard, Fig. 7A The example of illustrates a direct path unit. Figure 7B The example utilizes an optical reflector 87 at one end of a ferrule 85 for reflecting the optical signal and a circulator 84 located between the input multi-core fiber 80a and the output multi-core fiber 80b and in turn couples them to the intermediate multi-core fiber segment 80c.

[0103] Thus, the present technology provides artificial optical neuron units configured to operate in artificial neuron networks. The present technology allows for selective weighting of input signal portions as well as mixing and manipulation of optical signals. The present technology enables all-optical neuron networks, or at least nearly all-optical, where weight selection, signal processing and even certain training operations can be determined directly based on optical signals without conversion to electrical signals.

Claims

1. A processing node for use in an artificial neural network, the processing node comprising an input port suitable for receiving a first input optical signal and a second input optical signal and an optical spatial mixing device, the optical spatial mixing device being configured to receive the first input optical signal and the second input optical signal and to apply optical processing to the first input optical signal and the second input optical signal to provide output data indicative of a correlation between the first input optical signal and the second input optical signal.

2. The treatment structure according to claim 1, wherein: The first input optical signal and the second input optical signal are associated with optical signals propagating in corresponding first and second waveguides.

3. The treatment structure according to claim 1 or 2, wherein: The optical spatial mixing device comprises at least one optical reflective element configured to reflect light of the first input optical signal and the second input optical signal into a common spatial path, thereby enabling optical measurement of the spatial correlation between the first input optical signal and the second input optical signal.

4. The processing node according to claim 1 or 2 further comprises a decoherence unit, which is configured to reduce the spatial coherence of the first input optical signal and the second input optical signal upstream of the optical spatial mixing device.

5. The processing node according to claim 1 or 2 is configured to receive the first input optical signal and the second input optical signal through a first multi-core optical waveguide and a second multi-core optical waveguide, and the optical spatial mixing device is configured to provide interaction and enhanced mixing between the first input optical signal and the second input optical signal in the free space propagation of the optical signal.

6. An optical processing unit, comprising an optical gain unit, wherein the optical gain unit has an input face and an output face located in an optical path between an input multi-core optical waveguide and an output multi-core optical waveguide; the optical gain unit is exposed to external illumination to generate a holographic pattern within the optical gain unit, thereby selectively affecting the light transmission between the input multi-core optical waveguide and the output multi-core optical waveguide.

7. The optical processing unit according to claim 6, wherein: The holographic pattern within the optical gain unit is three-dimensional.

8. The optical processing unit according to claim 6 or 7, further comprising an input optical lens unit and an output optical lens unit, wherein the input optical lens unit and the output optical lens unit are respectively located in an optical path between the input multi-core optical waveguide and an input surface of the optical gain unit and in an optical path between an output surface of the optical gain unit and the output multi-core optical waveguide.

9. The optical processing unit according to claim 8, wherein: The input multi-core optical waveguide and the output multi-core optical waveguide are formed as one-dimensional waveguides, and the input optical lens unit is an astigmatic lens configured to guide input light to form a three-dimensional spatial pattern within the optical gain unit.

10. An optical processing unit comprising at least one optical input port for receiving a first optical signal and at least one additional input port for receiving a second additional input signal, the optical processing unit comprising a first optical waveguide segment, an interaction node, and a second optical waveguide segment, wherein: The first optical waveguide segment and the second optical waveguide segment have selected characteristics and lengths for separating optical signals passing through the first optical waveguide segment and the second optical waveguide segment according to wavelength or spatial frequency components, and the interaction node is configured to perform the following operations: receive a signal component of the first optical signal from the first optical waveguide segment, interact the signal component with the second additional input signal, and generate a multiplied signal component, and couple the multiplied signal component to the second optical waveguide segment for transforming the multiplied signal component, and provide an output signal indicating the interaction between the first optical signal and the second additional input signal.

11. The optical processing unit according to claim 10, wherein: The first optical waveguide segment and the second optical waveguide segment include a graded refractive index segment having a selected refractive index profile and length for separating an optical signal passing through the graded refractive index segment into spatial frequencies of the optical signal, thereby achieving spatial Fourier transform of an input signal.

12. An optical processing unit according to claim 11, comprising an arrangement of multiple optical waveguide cores formed by optical waveguides carrying gain material, and the second additional input signal provides selective pumping to the arrangement of multiple optical waveguide cores, thereby selectively causing components of the second additional input signal to interact with spatial components of the first optical signal.

13. The optical processing unit according to any one of claims 10 to 12, wherein: The first optical waveguide segment and the second optical waveguide segment include dispersive optical waveguides having a length selected to apply a Fourier transform to the optical signal with respect to wavelength components of the optical signal, and the interaction node includes a time modulator configured to receive data indicative of the second additional input signal and to modulate components of the first optical signal accordingly, thereby causing frequency components of the first optical signal and the second additional input signal to interact.

Images