A reservoir computing system and method based on a semiconductor laser
The hybrid optical-electrical reservoir computing system enhances semiconductor laser-based systems by integrating electrical signals to increase virtual nodes, addressing the trade-offs of accuracy, speed, cost, and size in real-time applications.
Patent Information
- Application Number
- CN202510513925.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-23
- Publication Date
- 2025-07-15
- Estimated Expiration
- 2045-04-23
AI Technical Summary
The existing reserve pool computing system cannot take into account data processing accuracy, data processing speed, cost, power consumption and volume, and is particularly insufficient in application scenarios with high real-time requirements.
The photoelectric hybrid reserve pool module is adopted to laser response to the original optical signal through a semiconductor laser, and the feedback loop and coupling module are used to divide the response signal into two optical signals for coupling and feedback. Combined with electrical signal acquisition, a rich reserve pool state is generated, the number of virtual nodes is increased, and the nonlinear characteristics of the system are enhanced.
Without changing the system hardware structure, data processing accuracy and efficiency are improved, hardware cost and power consumption are reduced, and it is suitable for application scenarios with high real-time requirements.
Smart Images

Figure CN120045818B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of reservoir computing, and in particular to a reservoir computing system and method based on semiconductor lasers. Background Art
[0002] In the context of the rapid development of information technology today, artificial intelligence is increasingly widely used, especially in the fields of data processing and pattern recognition. Traditional computing systems often struggle when faced with complex non-linear problems. Therefore, researchers have been constantly exploring new computing architectures to improve processing efficiency and accuracy. Neural networks, as a computing system that mimics the biological nervous system, have become the core for solving various intelligent tasks. Its basic unit - neurons form a complex network structure through connections between layers, and have achieved remarkable results in many applications such as robot path planning, image processing, and time series prediction. However, the training process of deep learning neural network computing systems requires a large amount of labeled data and computing resources, which will consume a large amount of human, material, and time costs.
[0003] To solve these problems, reservoir computing (RC) has gradually attracted attention as an emerging framework. It uses a series of physical components (such as multiple semiconductor lasers) to form a fixed, randomly connected dynamic computing system to process input signals, and trains the output layer through a simple linear regression method, enabling functions such as image classification, time series prediction, robot motion control, and path planning without a complex training process. Among them, semiconductor lasers have become one of the ideal devices for reservoir computing due to their advantages such as low power consumption and fast response speed. Taking a reservoir system based on semiconductor lasers as an example, when classifying an image, only the image to be classified needs to be encoded as an input optical signal and input into the reservoir. After the semiconductor laser receives the input optical signal, it will generate a response signal due to its own non-linear optical characteristics and optical feedback effect. These response signals contain feature information such as the space and gray scale of the image to be classified. The output module can obtain the output weights of the reservoir by training the output signals of the reservoir, and finally use the output weights to obtain the predicted category of the output image to be classified.
[0004] In a reservoir computing system based on semiconductor lasers, each semiconductor laser is a physical node that can respond to input signals. The number of nodes determines the amount of characteristic information that can be captured. For example, when performing time series prediction, the more nodes there are, the longer the historical information that can be recorded, and thus the more accurately the dynamic change law of the input data can be captured, improving the prediction accuracy of the system. In the prior art, the number of nodes is increased by increasing the number of components in the reservoir, thereby improving the data processing accuracy of the reservoir computing system. However, this not only increases the cost, hardware volume, and power consumption of the system, but also affects the data processing speed of the system, making the reservoir computing system unable to be used in application scenarios with high real-time requirements such as autonomous driving and medical monitoring.
[0005] In summary, the existing reservoir computing systems have problems in that they cannot balance data processing accuracy, data processing speed, cost, power consumption, and volume. Summary of the Invention
[0006] Therefore, the technical problem to be solved by the present invention is to overcome the problem that the existing reservoir computing systems cannot balance data processing accuracy, data processing speed, cost, power consumption, and volume.
[0007] To solve the above technical problem, the present invention provides a reservoir computing system based on semiconductor lasers, including:
[0008] An input module for generating an original optical signal;
[0009] An optoelectronic hybrid reservoir module, including:
[0010] A semiconductor laser for performing laser response to the original optical signal and the coupled optical signal transmitted through the feedback loop;
[0011] A feedback loop connected to the semiconductor laser for splitting the response signal output by the semiconductor laser into two optical signals, one as the response optical signal transmitted to the output module and the other as the feedback optical signal transmitted to the first coupling module; transmitting the coupled optical signal output by the first coupling module to the semiconductor laser;
[0012] A first coupling module connected to the input module and the feedback loop for coupling the original optical signal and the feedback optical signal and outputting a coupled optical signal;
[0013] An output module connected to the semiconductor laser and the feedback loop for collecting the response optical signal and the electrical signal of the semiconductor laser and obtaining the target output signal of the reservoir computing system based on the response optical signal and the electrical signal.
[0014] Preferably, the feedback loop includes:
[0015] An optical circulator, with its input end connected to the output end of a semiconductor laser and the output end of a first coupling module, and its output end connected to the input end of the semiconductor laser; it is used to receive the response signal output by the semiconductor laser and transmit the coupled optical signal output by the first coupling module to the semiconductor laser;
[0016] A second coupling module, with its input end connected to the output end of the optical circulator and its output end connected to the input end of a first variable optical attenuator and the input end of an output module; it is used to divide the response signal into two optical signals, one as the response optical signal transmitted to the output module and the other as the feedback optical signal transmitted to the first variable optical attenuator;
[0017] A first variable optical attenuator, used to adjust the feedback intensity of the feedback optical signal;
[0018] A signal delay transmission module, with its input end connected to the output end of the first variable optical attenuator and its output end connected to the input end of the first coupling module, used to perform delay processing on the feedback optical signal and transmit it to the first coupling module.
[0019] Preferably, the signal delay transmission module is a delay optical fiber or spatial light.
[0020] Preferably, the output module includes:
[0021] An optical signal collector, connected to the second coupling module, used to collect the response optical signal;
[0022] An electrical signal collector, connected to the semiconductor laser, used to collect the electrical signal of the semiconductor laser;
[0023] An adder, connected to the optical signal collector and the electrical signal collector, used to superimpose the response optical signal and the electrical signal to obtain an output signal;
[0024] A programmable gate array, connected to the adder, used to train the output of the optoelectronic hybrid reservoir module to obtain the output weights of the response optical signal and the electrical signal, so as to obtain the target output signal of the reservoir computing system.
[0025] Preferably, the electrical signal collector is a multimeter, used to collect the voltage signal of the semiconductor laser.
[0026] Preferably, the optical signal collector includes:
[0027] An optical isolator, connected to the second coupling module, used to transmit the response optical signal;
[0028] A photodetector, connected to the optical isolator, used to perform optoelectronic conversion on the response optical signal.
[0029] Preferably, the input module includes:
[0030] A driving laser for generating a laser signal;
[0031] A waveform generator for generating a mask signal;
[0032] A modulator, with its input end connected to the driving laser and the waveform generator, for modulating the laser signal based on the mask signal to generate an original optical signal.
[0033] Preferably, the input module further includes a second variable optical attenuator, with its input end connected to the output end of the modulator, for adjusting the injection intensity of the original optical signal.
[0034] Preferably, the optoelectronic hybrid reservoir module further includes a temperature controller, connected to the semiconductor laser, for controlling the central wavelength of the semiconductor laser.
[0035] The present invention also provides a reservoir computing method based on a semiconductor laser, which is implemented by using the above-mentioned reservoir computing system based on a semiconductor laser, and includes:
[0036] Using the input module to generate an original optical signal and transmit it to the first coupling module and the semiconductor laser, so that the semiconductor laser performs a laser response to the original optical signal;
[0037] Using the feedback loop to divide the response signal output by the semiconductor laser into two optical signals, one as the response optical signal is transmitted to the output module, and the other as the feedback optical signal is transmitted to the first coupling module;
[0038] Using the first coupling module to couple the original optical signal and the feedback optical signal, output the coupled optical signal and transmit it to the feedback loop;
[0039] Using the feedback loop to transmit the coupled optical signal to the semiconductor laser, so that the semiconductor laser performs a laser response to the coupled optical signal;
[0040] Using the output module to collect the response optical signal and the electrical signal of the semiconductor laser, and obtaining the target output signal of the reservoir computing system based on the response optical signal and the electrical signal.
[0041] The reservoir computing system based on a semiconductor laser provided by the present application has the following beneficial effects:
[0042] It is found in this application that when a semiconductor laser performs a laser response to an optical signal, physical processes such as electron transitions and carrier concentrations inside it will change, and thus its own electrical signal will change. Therefore, the electrical signal during the laser response of the semiconductor laser can also reflect the state of the system. Based on this, this application innovatively designs an optoelectronic hybrid reservoir module to replace the traditional reservoir module. When the optoelectronic hybrid reservoir module receives the original optical signal, the semiconductor laser performs a laser response to the original optical signal and outputs a response signal containing characteristic information. The response signal is divided into two optical signals by using a feedback loop. One is transmitted to the output module as a response optical signal, and the other is transmitted to the first coupling module as a feedback optical signal. The first coupling module couples the original optical signal containing the original information and the feedback optical signal containing the characteristic information to obtain a coupled optical signal integrating the original information and the characteristic information. The coupled optical signal is transmitted to the semiconductor laser again by using the feedback loop, so that the semiconductor laser performs a laser response to the coupled optical signal again to extract new characteristic information. Further, the output module simultaneously collects the response optical signal and the electrical signal of the semiconductor laser, reflects the state of the system after processing the input signal from different dimensions, generates rich reservoir states, enhances the non-linear characteristics of the system, improves the performance of the system when processing complex tasks, adds virtual nodes related to electrical signals on the basis of the original single optical signal node, thereby effectively increasing the number of virtual nodes without changing the hardware structure of the system, further enhancing the computing power and data processing accuracy of the system, and at the same time ensuring the characteristics of high data processing efficiency, low hardware cost, small volume and low power consumption of the system, greatly optimizing the system performance. Description of the Drawings
[0043] In order to make the content of the present invention easier to be clearly understood, the following further details the present invention according to specific embodiments of the present invention in conjunction with the drawings, where:
[0044] Figure 1 Schematic structural diagram of a reservoir computing system based on a semiconductor laser provided by this application;
[0045] Figure 2 Flowchart of a reservoir computing method based on a semiconductor laser provided by this application;
[0046] Figure 3 Schematic diagram of the prediction result when using the reservoir computing system provided by this application to process the Santa Fe time series prediction task; where, Figure 3 in (a) is a schematic diagram of the actual time series, Figure 3 in (b) is a schematic diagram of the predicted time series output by the reservoir computing system provided by this application, Figure 3 in (c) is a schematic diagram of the error between the actual result and the predicted result;
[0047] Description of the reference numerals in the drawings: 1. Input module; 11. Driving laser; 12. Waveform generator; 13. Modulator; 14. Second variable optical attenuator; 2. Optoelectronic hybrid reservoir module; 21. Semiconductor laser; 22. Feedback loop; 221. Optical circulator; 222. Second coupling module; 223. First variable optical attenuator; 224. Signal delay transmission module; 23. First coupling module; 24. Temperature controller; 3. Output module; 31. Optical signal collector; 311. Optical isolator; 312. Photoelectric detector; 32. Electrical signal collector; 33. Adder; 34. Programmable gate array. Detailed implementation manners
[0048] The present invention will be further described below in conjunction with the drawings and specific embodiments, so that those skilled in the art can better understand the present invention and be able to implement it, but the specific embodiments cited are not intended to limit the present invention.
[0049] Please refer to Figure 1 , Figure 1 , which is a schematic structural diagram of a reservoir computing system based on a semiconductor laser provided by the present application. The computing system specifically includes an input module 1, an optoelectronic hybrid reservoir module 2, and an output module 3.
[0050] The input module 1 is used to generate an original optical signal.
[0051] The optoelectronic hybrid reservoir module 2 includes a semiconductor laser 21, a feedback loop 22, and a first coupling module 23.
[0052] The semiconductor laser 21 is used to perform a laser response to the original optical signal and the coupled optical signal transmitted by the feedback loop 22.
[0053] The feedback loop 22 is connected to the semiconductor laser 21 and is used to divide the response signal output by the semiconductor laser 21 into two optical signals. One is transmitted to the output module 3 as a response optical signal, and the other is transmitted to the first coupling module 23 as a feedback optical signal; the coupled optical signal output by the first coupling module 23 is transmitted to the semiconductor laser 21.
[0054] The first coupling module 23 is connected to the input module 1 and the feedback loop 22 and is used to couple the original optical signal and the feedback optical signal, output a coupled optical signal and transmit it to the feedback loop 22. Specifically, the first coupling module 23 is an optical coupler.
[0055] Optionally, the original optical signal generated by the input module 1 can be input into the semiconductor laser 21 through the first coupling module 23 and the feedback loop 22, so that the semiconductor laser 21 can perform a laser response to it. The input module 1 can also be connected to the first coupling module 23 and the semiconductor laser 21 respectively, so that the original optical signal generated by the input module 1 is directly transmitted to the semiconductor laser 21.
[0056] In some embodiments, the optoelectronic hybrid reservoir module 2 further includes a temperature controller 24, which is connected to the semiconductor laser 21 and is used to control the central wavelength of the semiconductor laser 21, so as to control the frequency detuning between the semiconductor laser 21 and the input module 1.
[0057] The output module 3 is connected to the semiconductor laser 21 and the feedback loop 22, and is used to collect the response optical signal and the electrical signal of the semiconductor laser 21, and obtain the target output signal of the reservoir computing system based on the response optical signal and the electrical signal.
[0058] Since the characteristics such as the intensity and frequency of different original optical signals will cause the semiconductor laser to make different excitation responses and output optical signals with different characteristics such as intensity, wavelength and phase, in the prior art, the response optical signal when the semiconductor laser performs a laser response is collected to reflect the state performance of the computing system after processing the input signal. This application discovers that when the semiconductor laser performs a laser response to the original optical signal, physical processes such as internal electron transitions and carrier concentrations will change, and thus its own electrical signal will change. Therefore, the electrical signal when the semiconductor laser performs a laser response can also reflect the state of the system after processing the input signal.
[0059] Based on this, the present application innovatively designs an optoelectronic hybrid reservoir module to replace the traditional reservoir module. Specifically, when the optoelectronic hybrid reservoir module receives the original optical signal, the semiconductor laser performs a laser response to the original optical signal and outputs a response signal containing characteristic information. Using a feedback loop, the response signal is divided into two optical signals. One is transmitted as a response optical signal to the output module, and the other is transmitted as a feedback optical signal to the first coupling module. The first coupling module couples the original optical signal containing the original information and the feedback optical signal containing the characteristic information to obtain a coupled optical signal integrating the original information and the characteristic information. The feedback loop is used to transmit the coupled optical signal to the semiconductor laser again, enabling the semiconductor laser to perform a laser response to the coupled optical signal again and extract new characteristic information. Further, the output module simultaneously collects the response optical signal and the electrical signal of the semiconductor laser, reflecting the state of the system after processing the input signal from different dimensions, generating rich reservoir states, enhancing the nonlinear characteristics of the system, improving the performance of the system when processing complex tasks, adding virtual nodes related to electrical signals on the basis of the original single optical signal node, thus effectively increasing the number of virtual nodes without changing the hardware structure of the system, further enhancing the computing power and data processing accuracy of the system, while also ensuring the characteristics of high data processing efficiency, low hardware cost, small volume, and low power consumption of the system, and greatly optimizing the system performance.
[0060] Further, as Figure 1 shown in
[0061] The input end of the optical circulator 221 is connected to the output end of the semiconductor laser 21 and the output end of the first coupling module 23, and the output end is connected to the input end of the semiconductor laser 21; it is used to receive the response signal output by the semiconductor laser 21 and transmit the coupled optical signal output by the first coupling module 23 to the semiconductor laser.
[0062] The input end of the second coupling module 222 is connected to the output end of the optical circulator 221, and the output end is connected to the input end of the first variable optical attenuator 223 and the input end of the output module 3; it is used to divide the response signal into two optical signals, one is transmitted as a response optical signal to the output module 3, and the other is transmitted as a feedback optical signal to the first variable optical attenuator 223.
[0063] The first variable optical attenuator 223 is used to adjust the feedback intensity of the feedback optical signal.
[0064] The input end of the signal delay transmission module 224 is connected to the output end of the first variable optical attenuator 223, and the output end is connected to the input end of the first coupling module 23, which is used to delay the feedback optical signal and transmit it to the first coupling module 23.
[0065] Optionally, the signal delay transmission module 224 can be a delay optical fiber or free space light. Since the delay optical fiber has the advantages of small dispersion, large transmission bandwidth, and small attenuation, and is suitable for transmitting high-frequency signals, in some embodiments of the present application, a preset delay optical fiber is selected to transmit the feedback optical signal to the first coupling module 23. In other embodiments, free space light can also be used to transmit the feedback optical signal, and the present application does not limit this.
[0066] Furthermore, as Figure 1 shown in
[0067] The output module 3 includes an optical signal collector 31, an electrical signal collector 32, an adder 33, and a programmable gate array 34.
[0068] Specifically, in some embodiments of the present application, the optical signal collector 31 includes an optical isolator 311 and a photodetector 312. Among them, the optical isolator 311 is connected to the second coupling module 222 for transmitting the response optical signal; the photodetector 312 is connected to the optical isolator 311 for performing optoelectronic conversion on the response optical signal.
[0069] The electrical signal collector 32 is connected to the semiconductor laser 21 for collecting the electrical signal of the semiconductor laser 21.
[0070] Specifically, the voltage signal of the semiconductor laser 21 can reflect the system state. Therefore, the electrical signal collector 32 can be a multimeter for collecting the voltage signal of the semiconductor laser 21.
[0071] The adder 33 is connected to the optical signal collector 31 and the electrical signal collector 32 for superimposing the response optical signal and the electrical signal to obtain an output signal.
[0072] The programmable gate array 34 is connected to the adder 33 for training the output of the optoelectronic hybrid reservoir module 2 to obtain the output weights of the response optical signal and the electrical signal, thereby obtaining the target output signal of the reservoir computing system.
[0073] Exemplarily, after the photodetector 312 performs optoelectronic conversion on the response optical signal, an electrical signal is output , the electrical signal collector 32 collects the terminal voltage signal of the semiconductor laser 21 , and the two signals are superimposed by the adder 33 to form the output of the optoelectronic hybrid reservoir module 2 。The programmable gate array 34 trains the output of the optoelectronic hybrid reservoir module 2 using a linear regression algorithm or a ridge regression algorithm to obtain the output weights of the two signals, so as to obtain the target output signal of the reservoir computing system based on the two signals and their output weights.
[0074] As Figure 1 shown, in some embodiments of the present application, the input module 1 includes a driving laser 11, a waveform generator 12, and a modulator 13.
[0075] The driving laser 11 is used to generate a laser signal.
[0076] The waveform generator 12 is used to generate a mask signal.
[0077] The input end of the modulator 13 is connected to the driving laser 11 and the waveform generator 12, and is used to modulate the laser signal based on the mask signal to generate an original optical signal.
[0078] Specifically, the mask signal generated by the waveform generator 12 changes according to the requirements of the calculation scenario. The modulator 13 is a phase modulator or an intensity modulator, and the modulated original optical signal is obtained by multiplying the laser signal by the mask signal.
[0079] Optionally, in some embodiments of the present application, the input module 1 further includes a second variable optical attenuator 14, whose input end is connected to the output end of the modulator 13 and is used to adjust the injection intensity of the original optical signal.
[0080] The reservoir computing system based on a semiconductor laser provided by the present application can generate rich reservoir states by adjusting the feedback intensity of the feedback optical signal, the injection intensity of the original optical signal, and the frequency detuning between the semiconductor laser and the driving laser, and combining the time-division multiplexing technology. At the same time, the dual output mechanism of the feedback loop and optoelectronic hybrid can effectively increase the non-linear virtual nodes of the system without changing the hardware structure of the system, which can not only increase the number of captured feature information and improve the data processing accuracy, but also has the advantages of high data processing efficiency, low cost, and low power consumption.
[0081] Based on the reservoir computing system based on a semiconductor laser provided in the above embodiments, the embodiments of the present application also provide a reservoir computing method based on a semiconductor laser, as Figure 2 shown, the method specifically includes:
[0082] S10: Use the input module to generate an original optical signal and transmit it to the first coupling module and the semiconductor laser, so that the semiconductor laser can perform a laser response to the original optical signal.
[0083] S20: Divide the response signal output by the semiconductor laser into two optical signals using a feedback loop. One is transmitted to the output module as the response optical signal, and the other is transmitted to the first coupling module as the feedback optical signal.
[0084] S30: Couple the original optical signal and the feedback optical signal using the first coupling module, output the coupled optical signal, and transmit it to the feedback loop.
[0085] S40: Transmit the coupled optical signal to the semiconductor laser using the feedback loop so that the semiconductor laser can perform a laser response to the coupled optical signal.
[0086] S50: Collect the response optical signal and the electrical signal of the semiconductor laser using the output module, and obtain the target output signal of the reservoir computing system based on the response optical signal and the electrical signal.
[0087] The embodiment of the present application also applies the reservoir computing system based on a semiconductor laser provided in the above embodiment to the Santa Fe time series prediction task. Specifically, first, a reservoir computing system based on a semiconductor laser is simulated based on numerical simulation, and the obtained rate equations are as follows:
[0088] ,
[0089] ,
[0090] Among them, represents the complex amplitude of the electric field of the semiconductor laser; represents the carrier concentration of the semiconductor laser; represents the linewidth enhancement factor; represents the gain coefficient; represents the transparent carrier number; represents the gain saturation factor; represents the photon lifetime; represents the carrier lifetime in the active region, represents the injection current of the semiconductor laser; represents the feedback intensity of the semiconductor laser; represents the angular frequency of the semiconductor laser; represents the feedback time delay; represents the injection intensity of the driving laser; represents the optical intensity of the output light of the driving laser; represents the optical injection frequency detuning; , represents the input information stream after mask signal processing, where is the input corresponding to the task, is the mask signal, is the mask scaling factor; represents an imaginary number; represents the spontaneous emission factor; represents a noise signal with zero mean and variance of 0.
[0091] The values of each parameter in the simulation are as follows: ; ; ; ; ; ; ; ; ; ; . Set the virtual node interval during the simulation , the delay time is set to 1 ns, and the information processing rate is 1 Gbps.
[0092] Specifically, the specific steps for time series prediction using this system are as follows:
[0093] Step 1: The waveform generator generates a mask signal based on the time series to be predicted, drives the laser to generate a laser signal, and the modulator multiplies the mask signal and the laser signal to obtain the original optical signal modulated with the time series information to be predicted;
[0094] Step 2: Input the original optical signal into the optoelectronic hybrid reservoir module, use the semiconductor laser to extract the characteristic information of the original optical signal, and output a response signal containing the characteristic information;
[0095] Step 3: Use the feedback loop to divide the response signal into two paths, one path is used as the response optical signal containing the characteristic information and transmitted to the output module, and the other path is used as the feedback optical signal containing the characteristic information and transmitted to the first coupling module;
[0096] Step 4: The first coupling module couples the original optical signal and the feedback optical signal, outputs a coupled optical signal containing the characteristic information and the time series information to be predicted, and transmits the coupled optical signal to the feedback loop;
[0097] Step 5: The feedback loop transmits the coupled optical signal to the semiconductor laser, uses the semiconductor laser to extract the characteristic information of the coupled optical signal, and outputs a response signal containing the characteristic information;
[0098] Step 6: The output module collects the response optical signal and the electrical signal of the semiconductor laser to obtain the output of the optoelectronic hybrid reservoir module, trains the output to obtain the output weights of the response optical signal and the electrical signal, and obtains the predicted time series based on the response optical signal, the electrical signal and their output weights.
[0099] Such as Figure 3The figure shows a schematic diagram of the prediction results when the reservoir computing system provided by this application is used to process the Santa Fe time series prediction task; among them, Figure 3 in (a) is a schematic diagram of the actual time series, Figure 3 in (b) is a schematic diagram of the predicted time series output by the reservoir computing system provided by this application, Figure 3 in (c) is a schematic diagram of the error between the actual result and the predicted result. As can be seen from Figure 3 it, the predicted time series is very close to the actual time series, and the value of the calculated normalized mean square error NMSE is only 0.0057, which fully proves that the reservoir computing system based on semiconductor lasers designed in this application can better capture the dynamic timing characteristics of the input signal and has good data prediction performance.
[0100] Those skilled in the art should understand that the embodiments of this application can be provided as a method, a system, or a computer program product. Therefore, this application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0101] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of this application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.
[0102] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device implements the functions specified in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.
[0103] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable apparatus provide steps for realizing the functions specified in one process or multiple processes and / or blocks Figure 1 one process or multiple processes and / or blocks Figure 1 steps for realizing the functions specified in one block or multiple blocks.
[0104] Obviously, the above-described embodiments are merely examples given for clear illustration and are not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to exhaustively list all implementation manners here. And the obvious changes or modifications derived therefrom still fall within the protection scope of the present invention.
Claims
1. A reservoir computing system based on a semiconductor laser, characterized in that, Comprising: An input module for generating an original optical signal; An optoelectronic hybrid reservoir computing module, comprising: A semiconductor laser for laser responding to the original optical signal and the coupled optical signal transmitted by the feedback loop; A feedback loop connected to the semiconductor laser for dividing the response signal output by the semiconductor laser into two optical signals, one serving as a response optical signal and transmitted to the output module, and the other serving as a feedback optical signal and transmitted to the first coupling module; transmitting the coupled optical signal output by the first coupling module to the semiconductor laser; specifically including: An optical circulator, with its input end connected to the output end of the semiconductor laser and the output end of the first coupling module, and its output end connected to the input end of the semiconductor laser; for receiving the response signal output by the semiconductor laser and transmitting the coupled optical signal output by the first coupling module to the semiconductor laser; A second coupling module, with its input end connected to the output end of the optical circulator, and its output end connected to the input end of the first variable optical attenuator and the input end of the output module; for dividing the response signal into two optical signals, one serving as a response optical signal and transmitted to the output module, and the other serving as a feedback optical signal and transmitted to the first variable optical attenuator; A first variable optical attenuator for adjusting the feedback intensity of the feedback optical signal; A signal delay transmission module, with its input end connected to the output end of the first variable optical attenuator and its output end connected to the input end of the first coupling module, for delaying the feedback optical signal and transmitting it to the first coupling module; A first coupling module connected to the input module and the feedback loop for coupling the original optical signal and the feedback optical signal and outputting a coupled optical signal; An output module connected to the semiconductor laser and the feedback loop for collecting the response optical signal and the electrical signal of the semiconductor laser, and obtaining the target output signal of the reservoir computing system based on the response optical signal and the electrical signal.
2. The reservoir computing system based on a semiconductor laser according to claim 1, wherein The signal delay transmission module is a delay optical fiber or free space light.
3. The reservoir computing system based on a semiconductor laser according to claim 1, wherein The output module includes: An optical signal collector connected to the second coupling module for collecting the response optical signal; An electrical signal collector connected to the semiconductor laser for collecting the electrical signal of the semiconductor laser; An adder connected to the optical signal collector and the electrical signal collector for superimposing the response optical signal and the electrical signal to obtain an output signal; A programmable gate array connected to the adder for training the output of the optoelectronic hybrid reservoir computing module to obtain the output weights of the response optical signal and the electrical signal, thereby obtaining the target output signal of the reservoir computing system.
4. The reservoir computing system based on a semiconductor laser according to claim 3, characterized in that The electrical signal collector is a multimeter for collecting the voltage signal of the semiconductor laser.
5. The reservoir computing system based on a semiconductor laser according to claim 3, wherein The optical signal collector includes: An optical isolator connected to the second coupling module for transmitting the response optical signal; An optical detector connected to the optical isolator for performing optoelectronic conversion on the response optical signal.
6. The reservoir computing system based on a semiconductor laser according to claim 1, wherein The input module includes: A driving laser for generating a laser signal; A waveform generator for generating a mask signal; A modulator, with its input end connected to the driving laser and the waveform generator, for modulating the laser signal based on the mask signal to generate an original optical signal.
7. The reservoir computing system based on a semiconductor laser according to claim 6, wherein The input module further includes a second variable optical attenuator, with its input end connected to the output end of the modulator, for adjusting the injection intensity of the original optical signal.
8. The reservoir computing system based on a semiconductor laser according to claim 1, wherein The optoelectronic hybrid reservoir module further includes a temperature controller, connected to the semiconductor laser, for controlling the central wavelength of the semiconductor laser.
9. A reservoir computing method based on a semiconductor laser, characterized in that The method is implemented by using the reservoir computing system based on semiconductor lasers according to any one of claims 1 to 8, and includes: Generating an original optical signal by using an input module and transmitting it to a first coupling module and a semiconductor laser, so that the semiconductor laser performs a laser response to the original optical signal; Dividing the response signal output by the semiconductor laser into two optical signals by using a feedback loop, one of which is transmitted to an output module as a response optical signal, and the other is transmitted to the first coupling module as a feedback optical signal; specifically, it includes: Receiving the response signal output by the semiconductor laser by using an optical circulator; Dividing the response signal into two optical signals by using a second coupling module, one of which is transmitted to an output module as a response optical signal, and the other is transmitted to a first variable optical attenuator as a feedback optical signal; Adjusting the feedback intensity of the feedback optical signal by using the first variable optical attenuator; Performing a delay process on the feedback optical signal by using a signal delay transmission module and transmitting it to the first coupling module; Coupling the original optical signal and the feedback optical signal by using the first coupling module, outputting a coupled optical signal and transmitting it to the feedback loop; Transmitting the coupled optical signal to the semiconductor laser by using the optical circulator in the feedback loop, so that the semiconductor laser performs a laser response to the coupled optical signal; Collecting the response optical signal and the electrical signal of the semiconductor laser by using an output module, and obtaining the target output signal of the reservoir computing system based on the response optical signal and the electrical signal.
Citation Information
Patent Citations
Deep convolutional neural network system based on laser array and control method
CN114781602A
All-optical reserve pool parallel computing device
CN116257289A