Deep optical neural network model, architecture and system

Abstract

The invention relates to the technical field of optical computing, in particular to a deep optical neural network model, architecture and system. Wherein the model comprises an optical calculation module set, the connection mode between optical calculation modules in the optical calculation module set is determined by a target calculation task, each optical calculation module comprises an optical calculation matrix, and the number of rows and the number of columns of the optical calculation matrix are determined by the target calculation task. By adopting the scheme, a high-parameter-quantity complex intelligent model can be realized through the optical calculation module.

Description

Technical Field

[0001] The present disclosure relates to the field of optical computing technology, and in particular to a deep optical neural network model, architecture and system. Background Art

[0002] With the rapid development of artificial intelligence and scientific computing, the complexity and scale of computing needs are also increasing. However, the existing electronic computing technology is limited by Moore's Law, and its performance is gradually approaching saturation, making it difficult to effectively cope with the increasingly stringent requirements of large-scale complex algorithms on computing power and power consumption. Light has natural advantages such as high throughput and low latency in the propagation process. Optical computing technology that uses photons instead of electrons as computing carriers is seen as the key to breaking the existing computing bottleneck. Summary of the invention

[0003] The present disclosure aims to solve one of the technical problems in the related art at least to some extent.

[0004] To this end, the first objective of the present disclosure is to propose a deep optical neural network model to realize a complex intelligent model with a high number of parameters through an optical computing module.

[0005] The second objective of the present disclosure is to propose a deep optical neural network architecture.

[0006] The third objective of the present disclosure is to propose a deep optical neural network system.

[0007] To achieve the above objectives, the first embodiment of the present disclosure proposes a deep optical neural network model, including: An optical computing module set, wherein the connection mode between the optical computing modules in the optical computing module set is determined by the target computing task, and the optical computing module includes an optical computing matrix, and the number of rows and columns of the optical computing matrix is ​​determined by the target computing task.

[0008] Optionally, the target computing task is an optical residual neural network computing task, the optical computing module set includes multiple optical computing module subsets, the multiple optical computing module subsets are connected in series, the optical computing modules in the optical computing module subsets are connected in series, the optical computing matrices of the optical computing modules in the same optical computing module subset have the same number of rows and columns, the optical computing matrix is ​​used to implement at least one convolution kernel, and the input and output ends of the optical computing module subset are cross-layer connected within the optical computing module subset.

[0009] Optionally, the model further includes an electronic module, and the two optical computing module subsets connected in series are respectively a first optical computing module subset and a second optical computing module subset, and the first optical computing module subset and the second optical computing module subset are connected in series through the electronic module; wherein, The input end of the electronic module is respectively connected to the input end of the first optical computing module subset and the output end of the first optical computing module subset, and the output end of the electronic module is respectively connected to the input end of the second optical computing module subset and the output end of the second optical computing module subset. The electronic module is used to perform nonlinear calculation on the optical input signal and the optical output signal of the first optical computing module subset, and input the nonlinear calculation result as the optical input signal to the second optical computing module subset.

[0010] Optionally, the electronic module is further used for: The nonlinear calculation result is temporarily stored.

[0011] Optionally, the target computing task is an optical Transformer computing task, the model includes a plurality of Transformer decoding modules connected in series, the optical computing module set includes a plurality of optical computing module subsets, each of the Transformer decoding modules includes an optical computing module subset, and the optical computing modules in the optical computing module subsets are connected in series and / or in parallel to splice the optical computing matrix into a high-dimensional optical computing matrix corresponding to the optical Transformer computing task.

[0012] Optionally, the Transformer decoding module includes a self-attention submodule and a feedforward natural network submodule, the self-attention submodule includes the optical computing module subset, and the input end of the self-attention submodule is respectively connected to the output end of the self-attention submodule, the input end of the feedforward natural network submodule, and the output end of the feedforward natural network submodule.

[0013] Optionally, the self-attention submodule further includes a first optical linear layer and a second optical linear layer; wherein, The input end of the first optical linear layer is respectively connected to the output end of the second optical linear layer, the input end of the feedforward natural network sub-module and the output end of the feedforward natural network sub-module, the output end of the first optical linear layer is connected to the input end of the optical computing module subset, and the output end of the optical computing module subset is connected to the input end of the second optical linear layer.

[0014] Optionally, the feed-forward natural network submodule includes at least one third optical linear layer.

[0015] To achieve the above-mentioned purpose, the second aspect of the present disclosure proposes a deep optical neural network architecture, including: at least one deep optical neural network model shown in any one of the aforementioned first aspects.

[0016] To achieve the above-mentioned objectives, the third aspect of the present disclosure proposes a deep optical neural network system, comprising: at least one deep optical neural network model shown in any one of the aforementioned first aspects.

[0017] In summary, the deep optical neural network model, architecture and system provided by the present disclosure, through the reasonable reconstruction and structural combination of optical computing modules according to the target computing tasks, evolves matrix operations of different dimensions and scales, and performs vertical depth expansion, and can stack optical computing module slices into a neural network computing structure that can perform target computing tasks, and can effectively expand the depth and performance of the network. Compared with general optical neural networks, this deep optical neural network model can effectively stack the overall parameter amount (that is, as the parameter amount increases, the performance is effectively improved), and can realize a complex intelligent model with a high parameter amount.

[0018] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description or learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and / or additional aspects and advantages of the present disclosure will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which: Figure 1 A schematic diagram of the structure of a deep optical neural network model provided in an embodiment of the present disclosure; Figure 2 A schematic diagram of the architecture of an optical ResNet model provided in an embodiment of the present disclosure; Figure 3 A schematic diagram of an application of an optical ResNet model provided in an embodiment of the present disclosure; Figure 4 A schematic diagram of the architecture of an optical Transformer model provided in an embodiment of the present disclosure; Figure 5 A schematic diagram of an application of an optical Transformer model provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0020] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0021] With the rapid development of artificial neural network (ANN) technology, the performance and complexity of machine vision algorithms have been significantly improved, resulting in a continuous increase in the demand for high computing power. The bottleneck of traditional electronic computing architecture is its high power consumption and limited computing speed, which cannot fully meet the high parallel processing capabilities required for dynamic machine vision processing.

[0022] However, it is inherently difficult to achieve complex intelligent tasks with light. In order to cope with complex intelligent tasks, it is often necessary to expand the depth of neural networks. Extremely deep neural networks have stronger expressive power and can learn more complex nonlinear mappings. This makes them perform well when processing high-dimensional and complex data, such as image recognition, natural language processing and other tasks. But for most analog computing architectures, the lack of reconstruction and expansion capabilities restricts their deep stacking and also limits their ability to handle intelligent tasks. Existing optical computing models are often limited to simple full connection or convolution operations when implementing neural networks, and the relatively fixed structure of each computing unit also makes it difficult for these modules to be combined into large-scale, intelligent and effective deep optical neural networks. Therefore, even if the structures are superimposed in large quantities, they are difficult to show the intelligence of deep neural networks. This has led to the fact that most optical neural network work still remains in the processing of simple data sets and the application of idealized scenarios.

[0023] The present disclosure is described in detail below with reference to specific embodiments.

[0024] Figure 1 This is a schematic diagram of the structure of a deep optical neural network model provided by an embodiment of the present disclosure. Figure 1 As shown, the deep optical neural network model includes: An optical computing module set, wherein the connection mode between the optical computing modules in the optical computing module set is determined by the target computing task, the optical computing module includes an optical computing matrix, and the number of rows and columns of the optical computing matrix is ​​determined by the target computing task.

[0025] According to some embodiments, the number of rows and the number of columns of the optical calculation matrix correspond one-to-one with the input signal dimension and the input signal dimension in the optical calculation module, respectively. For example, the number of rows of the optical calculation matrix is ​​the same as the input signal dimension in the optical calculation module, and the number of columns of the optical calculation matrix is ​​the same as the output signal dimension in the optical calculation module; or, the number of rows of the optical calculation matrix is ​​the same as the output signal dimension in the optical calculation module, and the number of columns of the optical calculation matrix is ​​the same as the input signal dimension in the optical calculation module.

[0026] In some embodiments, the optical computing module can be reconstructed and expanded, the input signal dimension and the input signal dimension can be adjusted according to the target computing task, and the number of rows and columns of the optical computing matrix can also be adjusted according to the target computing task.

[0027] According to some embodiments, the target computing task refers to a task that requires a deep optical neural network model to perform computing, including but not limited to tasks such as image classification and natural language processing.

[0028] It should be noted that, due to the flexible reconstruction characteristics of the optical computing module, the present disclosure can successfully implement the special structure of each layer of the model in the design of the deep optical neural network, rather than simply splicing fixed-form computing units. This is also reflected in the scaling law of the model. As the scale of the network parameters increases, the performance of the deep optical neural network model gradually improves when dealing with complex image classification and language modeling problems. For example, in the language modeling task, when the network parameter amount increases from 5M, 30M, 60M, 0.117B to 0.345B parameters, the loss function decreases from 4.91 to 3.04, and the accuracy rate increases from 28.58% to 43.92%.

[0029] It is easy to understand that by rationally reconstructing and structurally combining the optical computing modules according to the target computing tasks, evolving matrix operations of different dimensions and scales, and performing vertical depth expansion, the optical computing module slices can be stacked into a neural network computing structure that can perform the target computing tasks, and can effectively expand the depth and performance of the network. Compared with general optical neural networks, this deep optical neural network model can effectively stack the overall parameter volume (that is, effectively improve performance as the parameter volume increases), and can realize complex intelligent models with high parameter volume.

[0030] Optionally, the target computing task is an optical residual neural network (ResNet) computing task, the optical computing module set includes multiple optical computing module subsets, the multiple optical computing module subsets are connected in series, the optical computing modules in the optical computing module subsets are connected in series, the optical computing matrices of the optical computing modules in the same optical computing module subset have the same number of rows and columns, the optical computing matrix is ​​used to implement at least one convolution kernel, and the input and output ends of the optical computing module subset are cross-layer connected within the optical computing module subset.

[0031] According to some embodiments, a ResNet computing task refers to a computing task that needs to be performed using ResNet.

[0032] In some embodiments, the number of rows and columns of the optical computing matrix of the optical computing modules in different optical computing module subsets may be the same or different, and may be determined by grouping the optical computing module set according to the target computing task.

[0033] For example, Figure 2 The schematic diagram of the architecture of an optical ResNet model provided by the embodiment of the present disclosure; wherein the boxes represent optical computing modules, 64 / 64, 128 / 128, 256 / 256, 512 / 512 represent the number of rows and columns of the optical computing matrix, and 3×3, 3×3 / 2 represent the dimensions of the convolution kernel. Figure 2 As shown in the figure, by deeply connecting multiple optical computing modules with different input and output scales, multiple 3×3 convolution kernels can be implemented inside each optical computing module, and then these optical computing modules are grouped, and cross-layer connections from input to output are made directly within the subset of optical computing modules. Therefore, each optical computing module contains the grouping and summation of the calculation results of multiple convolution kernels. By merging and calculating the outputs of the optical computing modules and then passing them to a similar structure of the next layer, convolution layers and deep networks can be implemented, thereby realizing optical ResNet.

[0034] According to some embodiments, the optical ResNet model further includes an electronic module, the two optical computing module subsets connected in series are respectively a first optical computing module subset and a second optical computing module subset, and the first optical computing module subset and the second optical computing module subset are connected in series through the electronic module; wherein, The input end of the electronic module is respectively connected to the input end of the first optical computing module subset and the output end of the first optical computing module subset, and the output end of the electronic module is respectively connected to the input end of the second optical computing module subset and the output end of the second optical computing module subset. The electronic module is used to perform nonlinear calculation on the optical input signal and the optical output signal of the first optical computing module subset, and input the nonlinear calculation result as the optical input signal to the second optical computing module subset.

[0035] In some embodiments, the representation capability of the optical ResNet model can be improved by using electronic modules to perform optoelectronic nonlinear operations between subsets of optical computing modules.

[0036] In some embodiments, the electronic module can also be used to temporarily store the nonlinear calculation results. Therefore, when the electronic module performs nonlinear calculation on the optical input signal and the optical output signal of the first optical calculation module subset, the optical output signal temporarily stored in the previous electronic module can be directly received.

[0037] In some embodiments, when the electronic module performs nonlinear calculations, the optical output signals may be read out through photoelectric conversion (ADC) and then combined for calculation.

[0038] In some embodiments, the electronic module may be implemented, for example, as a Field Programmable Gate Array (FPGA).

[0039] Take a scenario as an example. Figure 3 This is a schematic diagram of an application of an optical ResNet model provided by an embodiment of the present disclosure. Figure 3 As shown in the figure, the optical ResNet model is formed by stacking more than one hundred layers of optical computing modules. In the image classification task, the structure and scale of Resnet can be used to accurately identify and classify images in various complex scenes. Whether it is common objects in daily life or specific images in professional fields, accurate classification results can be obtained, which can reflect the intelligence of deep optical networks.

[0040] Optionally, the target computing task is an optical Transformer computing task, the model includes multiple Transformer decoding modules connected in series, the optical computing module set includes multiple optical computing module subsets, each Transformer decoding module includes an optical computing module subset, and the optical computing modules in the optical computing module subsets are connected in series and / or in parallel to splice the optical computing matrix into a high-dimensional optical computing matrix corresponding to the optical Transformer computing task.

[0041] According to some embodiments, a Transformer computing task refers to a computing task that needs to be performed using a Transformer.

[0042] According to some embodiments, the Transformer decoding module includes a self-attention submodule and a feedforward natural network submodule, the self-attention submodule includes a subset of optical computing modules, and the input end of the self-attention submodule is respectively connected to the output end of the self-attention submodule, the input end of the feedforward natural network submodule, and the output end of the feedforward natural network submodule.

[0043] In some embodiments, the self-attention submodule further includes a first optical linear layer and a second optical linear layer; wherein, The input end of the first optical linear layer is respectively connected to the output end of the second optical linear layer, the input end of the feedforward natural network submodule and the output end of the feedforward natural network submodule, the output end of the first optical linear layer is connected to the input end of the optical computing module subset, and the output end of the optical computing module subset is connected to the input end of the second optical linear layer.

[0044] In some embodiments, at least one third optical linear layer is included in the feed-forward natural network submodule.

[0045] In some embodiments, the first optical linear layer, the second optical linear layer, and the third optical linear layer may be implemented by an electronic computing module, which may be, for example, an electronic module similar to an FPGA.

[0046] It should be noted that by using the first optical linear layer, the second optical linear layer and the third optical linear layer to perform optoelectronic nonlinear operations between subsets of optical computing modules, the representation capability of the optical Transformer model can be improved.

[0047] For example, Figure 4 The following is a schematic diagram of the architecture of an optical Transformer model provided by an embodiment of the present disclosure. Figure 4 As shown in the figure, by reconstructing each optical computing module and connecting them in parallel and cascade in a certain number, the smaller matrices can be spliced ​​into a high-dimensional optical computing matrix (for example, a 128×128 matrix is ​​split into 128 128×1 matrices, each small matrix can be realized by an optical computing module, and the output of the optical computing module is arranged together to form a 128-dimensional vector), and then an optical linear layer that realizes nonlinearity and hierarchical connection is introduced at the output end, and the optical Transformer is realized by connecting them backward layer by layer. This process can not only improve the representation ability of the optical Transformer model, but also effectively use the advantages of optical computing to accelerate complex network operations.

[0048] Take a scenario as an example. Figure 5 This is a schematic diagram of an application of an optical Transformer model provided by an embodiment of the present disclosure. Figure 5 As shown, by applying the present disclosure to the fields of language processing and text-generated images (text-generated images), a transformer-based optical intelligence model with billions of parameters can be successfully constructed. In the text-generated image task, the optical Transformer model can generate an image that highly matches the description content based on the input text description, providing new possibilities for creative design and content creation.

[0049] In summary, the model provided in this embodiment can realize complex neural network structures such as optical ResNet and optical Transformer through a unique combination of reconfigurable optical computing modules, and can successfully solve the problem that the existing optical network is small in depth and scale, and the performance is difficult to effectively improve with the increase in the number of parameters. This innovation not only promotes the development of the field of optical computing, but also provides new ideas and directions for cutting-edge technologies such as artificial intelligence and deep learning. It has broad application prospects and is expected to bring new opportunities for high-performance intelligent computing in the post-Moore era and real-time analysis and control of transient scientific phenomena, and can be applied to unmanned systems, autonomous driving, ultrafast science and other fields.

[0050] In order to implement the above embodiments, the present disclosure also proposes a deep optical neural network architecture, including: a deep optical neural network model provided by at least one of the above embodiments.

[0051] In order to implement the above embodiments, the present disclosure also proposes a deep optical neural network system, including: at least one deep optical neural network model provided by the above embodiments.

[0052] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this disclosure shall comply with the relevant laws and regulations and shall not violate public order and good morals.

[0053] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.

[0054] The present disclosure anticipates providing implementation schemes for users to selectively block the use or access of personal information data. That is, the present disclosure anticipates providing hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, risks can be minimized by limiting data collection and deleting the data. In addition, when applicable, such personal information is de-identified to protect the privacy of the user.

[0055] The acquisition, transmission, storage, use, and processing of data in the technical solution disclosed in this disclosure are in compliance with the relevant provisions of national laws and regulations.

[0056] It should be noted that in the embodiments of the present disclosure, certain software, components, models and other existing solutions in the industry may be mentioned, which should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0057] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they contradict each other.

[0058] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0059] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present disclosure belong.

[0060] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or otherwise processing in a suitable manner if necessary, and then stored in a computer memory.

[0061] It should be understood that the various parts of the present disclosure can be implemented in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0062] A person skilled in the art may understand that all or part of the steps in the above-mentioned embodiment method may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0063] In addition, each functional unit in each embodiment of the present disclosure may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0064] The storage medium mentioned above may be a read-only memory, a disk or an optical disk, etc. Although the embodiments of the present disclosure have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present disclosure. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present disclosure.

Claims

1. A deep optical neural network model, characterized in that: include: An optical computing module set, wherein the connection mode between the optical computing modules in the optical computing module set is determined by the target computing task, and the optical computing module includes an optical computing matrix, and the number of rows and columns of the optical computing matrix is ​​determined by the target computing task.

2. The model according to claim 1, characterized in that The target computing task is an optical residual neural network computing task, the optical computing module set includes multiple optical computing module subsets, the multiple optical computing module subsets are connected in series, the optical computing modules in the optical computing module subsets are connected in series, the optical computing matrices of the optical computing modules in the same optical computing module subset have the same number of rows and columns, the optical computing matrix is ​​used to implement at least one convolution kernel, and the input and output ends of the optical computing module subset are cross-layer connected inside the optical computing module subset.

3. The model according to claim 2, characterized in that The model further includes an electronic module, and the two optical computing module subsets connected in series are respectively a first optical computing module subset and a second optical computing module subset, and the first optical computing module subset and the second optical computing module subset are connected in series through the electronic module; wherein, The input end of the electronic module is respectively connected to the input end of the first optical computing module subset and the output end of the first optical computing module subset, and the output end of the electronic module is respectively connected to the input end of the second optical computing module subset and the output end of the second optical computing module subset. The electronic module is used to perform nonlinear calculation on the optical input signal and the optical output signal of the first optical computing module subset, and input the nonlinear calculation result as the optical input signal to the second optical computing module subset.

4. The model according to claim 3, characterized in that The electronic module is also used for: The nonlinear calculation result is temporarily stored.

5. The model according to claim 1, characterized in that The target computing task is an optical Transformer computing task, the model includes a plurality of Transformer decoding modules connected in series, the optical computing module set includes a plurality of optical computing module subsets, each of the Transformer decoding modules includes one of the optical computing module subsets, and the optical computing modules in the optical computing module subsets are connected in series and / or in parallel to splice the optical computing matrix into a high-dimensional optical computing matrix corresponding to the optical Transformer computing task.

6. The model according to claim 5, characterized in that The Transformer decoding module includes a self-attention submodule and a feedforward natural network submodule, the self-attention submodule includes the optical computing module subset, and the input end of the self-attention submodule is respectively connected to the output end of the self-attention submodule, the input end of the feedforward natural network submodule, and the output end of the feedforward natural network submodule.

7. The model according to claim 6, characterized in that The self-attention submodule further includes a first optical linear layer and a second optical linear layer; wherein, The input end of the first optical linear layer is respectively connected to the output end of the second optical linear layer, the input end of the feedforward natural network sub-module and the output end of the feedforward natural network sub-module, the output end of the first optical linear layer is connected to the input end of the optical computing module subset, and the output end of the optical computing module subset is connected to the input end of the second optical linear layer.

8. The model according to claim 6, characterized in that The feed-forward natural network submodule includes at least one third optical linear layer.

9. A deep optical neural network architecture, characterized in that: include: A deep optical neural network model as described in any one of claims 1 to 8.

10. A deep optical neural network system, characterized in that: include: A deep optical neural network model as described in any one of claims 1 to 8.

Images