Method, device and system for marking data labels

Abstract

A method, device and system (10) for labeling data, relating to the field of computers, effectively improving the efficiency of labeling data. The method is applied to a computing device (20), comprising: determining a target control parameter based on first labeled data and first unlabeled data; sending the target control parameter to a quantum annealing device (30), the target control parameter being used to instruct the quantum annealing device (30) to perform quantum annealing to obtain a target annealing result, wherein the target annealing result includes the state of quanta in a quantum group (51, 52, 53, 54, 55, 56, 57, 58) corresponding to the first unlabeled data, and the state of quanta in the quantum group (51, 52, 53, 54, 55, 56, 57, 58) corresponding to the first unlabeled data is used to characterize the label of the first unlabeled data.

Description

Technical Field

[0001] The present application relates to the field of computers, and in particular to a method, device, and system for marking data tags. Background Art

[0002] When analyzing data, we often label the data. For example, if a computer can extract the characteristics of a picture of an orange: a round shape and an orange color, it can label the picture with the label "orange."

[0003] In the prior art, to improve the efficiency of data labeling, a label classifier can be generated by using quantum annealing (QA) to label the data. The quantum annealing method is usually used to find the global optimal solution in a discrete space, which has the advantage of high speed.

[0004] Specifically, the quantum annealing device performs supervised learning using quantum annealing on a large amount of labeled data with m (m is a positive integer greater than or equal to 1) labels, obtaining m label classifiers. Each classifier is used to label each label. Therefore, when labeling unlabeled data X using a label classifier, the computing device needs to perform 1 to m-1 calculations using these m label classifiers to obtain the label for data X.

[0005] As can be seen, the existing method of using quantum annealing to generate label classifiers for data labeling suffers from the following problems: First, this method requires a large amount of labeled data to generate the label classifiers, resulting in low accuracy when the amount of labeled data is small. Second, each classifier can only assign one label. When there are m types of labels, the computing device needs to perform 1 to m-1 calculations using m classifiers to obtain a label for the unlabeled data. Therefore, data labeling efficiency is low. Summary of the Invention

[0006] The present application provides a method, device and system for labeling data tags, which effectively improves the efficiency of labeling data tags.

[0007] To achieve the above objectives, this application provides the following technical solutions:

[0008] In a first aspect, the present application provides a method for labeling data, which is applied to a computing device. The method includes: determining a target control parameter based on first labeled data and first unlabeled data; wherein the target control parameter includes: a longitudinal field applied to the quanta in the quantum group corresponding to the first unlabeled data; or, the target control parameter includes: a longitudinal field applied to the quanta in the quantum group corresponding to the first labeled data, and a coupling effect applied between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data. Then, the computing device sends the target control parameter to a quantum annealing device, and the target control parameter is used to instruct the quantum annealing device to perform quantum annealing to obtain a target annealing result. Here, the target annealing result includes the state of the quanta in the quantum group corresponding to the first unlabeled data, and the state of the quanta in the quantum group corresponding to the first unlabeled data is used to characterize the label of the first unlabeled data.

[0009] The target control parameters of the quantum annealing device during quantum annealing are determined based on the labeled data and the unlabeled data, and the label of the unlabeled data is obtained based on the annealing results after the quantum annealing is completed. In this way, the method for labeling data labels provided by the present application does not require the generation of a classifier based on a large amount of labeled data in advance to label the unlabeled data, thereby reducing the amount of labeled data required when labeling the unlabeled data based on the labeled data and improving the efficiency of labeling the unlabeled data. In addition, if the target control parameters include the longitudinal field applied to the quantum in the quantum group corresponding to the first unlabeled data, that is, the computing device does not need to allocate a corresponding quantum group for the labeled data, thereby saving quantum resources and reducing the hardware implementation complexity during quantum annealing.

[0010] In combination with the first aspect, in a possible design method, the above-mentioned "determining the target control parameters based on the first labeled data and the first unlabeled data" includes: determining the target control parameters based on the similarity between the first labeled data and the first unlabeled data, and the encoding result of the label of the first labeled data.

[0011] In conjunction with the first aspect, in another possible design, if the computing device includes at least two labeled data, and the at least two labeled data include first labeled data, in this case, the above-mentioned "determining the target control parameter based on the similarity between the first labeled data and the first unlabeled data, and the encoding result of the label of the first labeled data" includes: determining the target control parameter based on the similarity between each labeled data of the at least two labeled data and the first unlabeled data, and the encoding result of the labels of the at least two labeled data.

[0012] In conjunction with the first aspect, in another possible design, when the at least two labeled data items have at least two labels, the method for labeling data further includes: sorting the labels of the at least two labeled data items based on the similarity between each pair of the at least two labels to obtain a first sequence. Then, encoding the labels in the first sequence using Gray coding to obtain encoding results of the at least two labels. Here, the encoding results of the at least two labels include the encoding result of the label of the first labeled data item.

[0013] After the computing device sorts the labels by similarity, it uses Gray coding to encode the labels of the data, which can make the encoding results of two labels with close similarity close, thereby improving the accuracy of labeling data after quantum annealing.

[0014] In combination with the first aspect, in another possible design method, if the encoding result of the label of any data in the computing device is represented by N (N is an integer greater than or equal to 1) bits, here, any data in the computing device can be the above-mentioned first labeled data, or the above-mentioned first unlabeled data, or any other labeled data or unlabeled data. In this case, "determining the target control parameter based on the similarity between the first labeled data and the first unlabeled data, and the encoding result of the label of the first labeled data" includes: determining the nth control parameter based on the similarity between the first labeled data and the first unlabeled data, and the nth (1≤n≤N, n is an integer) bit in the encoding result of the label of the first labeled data; wherein the target control parameter includes the 1st control parameter to the Nth control parameter. In this case, the above-mentioned “the computing device sends the target control parameter to the quantum annealing device” includes: the computing device sends the nth control parameter to the quantum annealing device, the nth control parameter is used to instruct the quantum annealing device to perform quantum annealing to obtain an nth annealing result, the nth annealing result includes the state of the quantum in the quantum group corresponding to the first unlabeled data; wherein, the first annealing result to the Nth annealing result are collectively used to represent the label of the first unlabeled data.

[0015] For one of the N bits representing a tag, the computing device sends the control parameter corresponding to that bit to the quantum annealing device, and by sending the control parameter N times, N annealing results are obtained. In other words, this application reduces the number of quanta required by the quantum annealing device during each annealing operation through time-division multiplexing, thereby conserving quantum resources and reducing the hardware implementation complexity during quantum annealing.

[0016] In conjunction with the first aspect, in another possible design, if the computing device includes at least two unlabeled data, and the at least two unlabeled data include first unlabeled data, in this case, the target control parameter also includes: a coupling effect applied between quantum groups corresponding to each pair of unlabeled data in the at least two unlabeled data. The target annealing result also includes: the states of quanta in the quantum groups corresponding to the unlabeled data in the at least two unlabeled data, excluding the first unlabeled data.

[0017] In combination with the first aspect, in another possible design, if the computing device includes Q (Q is an integer greater than or equal to 3) data, and the Q data include first labeled data and first unlabeled data, in this case, the above-mentioned method of marking data labels also includes: obtaining the similarity between the qth (1≤q≤Q, where q is an integer) data in the Q data and the data other than the qth data in the Q data. Then, based on the obtained similarity, the candidate data set corresponding to the qth data is determined. Here, the candidate data set corresponding to the qth data includes: data in the Q data whose similarity with the qth data is greater than or equal to a threshold, and the qth data has a connection relationship with each data in the candidate data set. In this case, the above-mentioned "determining the target control parameter based on the first labeled data and the first unlabeled data" includes: determining the target control parameter based on the similarity between the data in the Q data that has a connection relationship with the unlabeled data, and the encoding result of the label of the labeled data in the Q data.

[0018] It can be seen that for a piece of data, the present application selects data with a similarity value greater than or equal to a threshold as candidate data, or selects data with a similarity value ranked higher than the data as candidate data. Through these two data screening methods, the present application can reduce the number of data that has a connection relationship with the data, thereby reducing the number of data that need to be assigned quantum groups, thereby saving quantum resources and reducing the hardware implementation complexity during quantum annealing.

[0019] In a second aspect, the present application provides a method for marking data tags, which is applied to a quantum annealing device. The method includes: receiving a target control parameter sent by a computing device; wherein the target control parameter is determined based on the first labeled data and the first unlabeled data. The target control parameter includes: a longitudinal field applied to the quanta in the quantum group corresponding to the first unlabeled data; or, the target control parameter includes: a longitudinal field applied to the quanta in the quantum group corresponding to the first labeled data, and a coupling effect applied between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data. Then, the quantum annealing device performs quantum annealing according to the received target control parameter, obtains a target annealing result, and sends the target annealing result to the computing device. Here, the target annealing result includes the state of the quanta in the quantum group corresponding to the first unlabeled data, and the state of the quanta in the quantum group corresponding to the first unlabeled data is used to characterize the label of the first unlabeled data.

[0020] In conjunction with the second aspect, in one possible design, if the computing device includes at least two unlabeled data, and the at least two unlabeled data include first unlabeled data, the target control parameter further includes: a coupling effect applied between quantum groups corresponding to each pair of unlabeled data in the at least two unlabeled data; the at least two unlabeled data include the first unlabeled data. The target annealing result further includes: the states of quanta in the quantum groups corresponding to the unlabeled data in the at least two unlabeled data, excluding the first unlabeled data.

[0021] In conjunction with the second aspect, in another possible design, the aforementioned "the quantum annealing device performs quantum annealing according to the received target control parameters to obtain a target annealing result" includes: performing quantum annealing according to the nth (1≤n≤N, n is an integer, and N is an integer greater than or equal to 1) control parameter to obtain the nth annealing result. Here, the nth control parameter is used to indicate the longitudinal field of the quantum corresponding to the nth coding bit of the label of the first unlabeled data; or, the nth control parameter is used to indicate the longitudinal field of the quantum corresponding to the nth coding bit of the label of the first labeled data, and the coupling between the quantum corresponding to the nth coding bit in the coding results of the labels of the first labeled data and the first unlabeled data. When the label of any data has N coding bits, the target control parameters include the 1st to Nth control parameters. Here, any data can be the aforementioned first labeled data, the aforementioned first unlabeled data, or any other labeled data or unlabeled data. In addition, the above-mentioned “nth annealing result” includes the state of the quantum in the quantum group corresponding to the first unlabeled data; the first annealing result to the Nth annealing result are commonly used to characterize the label of the first unlabeled data.

[0022] In a third aspect, the present application provides a computing device that can be used to execute any of the methods provided in the first aspect.

[0023] In one possible design, the present application may divide the computing device into functional modules according to any of the methods provided in the first aspect above. For example, each functional module may be divided according to each function, or two or more functions may be integrated into one processing module. Exemplarily, the present application may divide the computing device into a determination unit and a sending unit, etc. according to the function. The description of the possible technical solutions and beneficial effects executed by each of the functional modules divided above can refer to the technical solutions provided by the first aspect or its corresponding possible design, and will not be repeated here.

[0024] In another possible design, the computing device includes: a memory and one or more processors, the memory and the processors being coupled. The memory is configured to store computer program code, the computer program code comprising computer instructions that, when executed by the computing device, cause the computing device to perform the data tagging method as described in the first aspect and any possible design thereof.

[0025] In a fourth aspect, the present application provides a quantum annealing device, which can be used to perform any of the methods provided in the second aspect above.

[0026] In one possible design, the present application can divide the quantum annealing device into functional modules according to any of the methods provided in the second aspect above. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. Exemplarily, the present application can divide the quantum annealing device into a receiving unit, a quantum annealing unit, and a transmitting unit, etc. according to the function. The description of the possible technical solutions and beneficial effects executed by each of the above-mentioned divided functional modules can refer to the technical solutions provided in the second aspect or its corresponding possible design, and will not be repeated here.

[0027] In another possible design, the quantum annealing device includes a memory and one or more processors, the memory and the processors being coupled. The memory is used to store computer program code, which includes computer instructions. When the computer instructions are executed by the quantum annealing device, the quantum annealing device performs the data tagging method described in the second aspect and any possible design thereof.

[0028] In a fifth aspect, the present application provides a quantum processor configured to execute the method for labeling data tags as described in the second aspect and any possible design thereof. The description of the possible technical solutions and beneficial effects implemented by the quantum processor can refer to the technical solutions provided in the first aspect or its corresponding possible designs, and will not be repeated here.

[0029] Specifically, the quantum processor may include a control circuit unit and a quantum unit, wherein the quantum unit includes one or more quanta. The control circuit unit is configured to receive target control parameters sent by a computing device and, based on the target control parameters, set a longitudinal field applied to quanta in the quantum group corresponding to the first unlabeled data; alternatively, set a longitudinal field applied to quanta in the quantum group corresponding to the first labeled data, and a coupling effect applied between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data. The quantum unit includes quanta in the quantum group corresponding to the first unlabeled data, or includes quanta in the quantum group corresponding to the first labeled data and quanta in the quantum group corresponding to the first unlabeled data.

[0030] In a sixth aspect, the present application provides a system for labeling data tags, comprising a computing device and a quantum annealing device. The computing device can be used to execute the method for labeling data tags as described in the first aspect and any possible design thereof; the quantum annealing device can be used to execute the method for labeling data tags as described in the second aspect and any possible design thereof. The possible technical solutions implemented by the system for labeling data tags, as well as the description of the corresponding beneficial effects, can be referenced to the technical solutions provided in the first aspect, the second aspect, or the corresponding possible designs, and will not be repeated here.

[0031] In a seventh aspect, the present application provides a chip system, which is applied to a computing device; the chip system includes one or more interface circuits and one or more processors. The interface circuits and processors are interconnected via circuits; the interface circuits are configured to receive signals from a memory of the computing device and send signals to the processors, the signals including computer instructions stored in the memory; when the processors execute the computer instructions, the computing device executes the method for tagging data as described in the first aspect and any possible design thereof.

[0032] In an eighth aspect, the present application provides a chip system, which is applied to a quantum annealing device; the chip system includes one or more interface circuits and one or more quantum processors. The interface circuits and the quantum processors are interconnected via circuits; the interface circuits are configured to receive signals from a memory of the quantum annealing device and send signals to the quantum processor, the signals including computer instructions stored in the memory; when the quantum processor executes the computer instructions, the quantum annealing device executes the method for labeling data tags as described in the second aspect and any possible design thereof.

[0033] In a ninth aspect, the present application provides a computer-readable storage medium comprising computer instructions, which, when executed on a computing device, enables the computing device to implement the method for marking data tags as described in the first aspect and any possible design thereof.

[0034] In a tenth aspect, the present application provides a computer-readable storage medium comprising computer instructions. When the computer instructions are executed on a quantum annealing device, the quantum annealing device implements the method for marking data tags as described in the second aspect and any possible design thereof.

[0035] In an eleventh aspect, the present application provides a computer program product, which, when executed on a computing device, enables the computing device to execute the method for marking data tags as described in the first aspect and any possible design thereof.

[0036] In a twelfth aspect, the present application provides a computer program product, characterized in that when the computer program product runs on a quantum annealing device, the quantum annealing device executes the method for marking data tags as described in the second aspect and any possible design thereof.

[0037] For the specific descriptions of the second to twelfth aspects and their various implementations in this application, reference can be made to the detailed descriptions in the first aspect and its various implementations; and for the beneficial effects of the second to tenth aspects and their various implementations, reference can be made to the analysis of the beneficial effects in the first aspect and its various implementations, which will not be repeated here.

[0038] In this application, the names of the computing device and quantum annealing device do not limit the devices or functional modules themselves. In actual implementation, these devices or functional modules may appear with other names. As long as the functions of each device or functional module are similar to those of this application, they are within the scope of the claims of this application and their equivalents.

[0039] These and other aspects of the present application will become more readily apparent from the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic diagram of a system for marking data tags provided in an embodiment of the present application;

[0041] Figure 2 A schematic diagram of the hardware structure of a computing device provided in an embodiment of the present application;

[0042] Figure 3 A schematic diagram of the hardware structure of a quantum annealing device provided in an embodiment of the present application;

[0043] Figure 4 A schematic diagram of a quantum state provided in an embodiment of the present application;

[0044] Figure 5 A schematic diagram of a quantum architecture provided in an embodiment of the present application Figure 1 ;

[0045] Figure 6a A flowchart of a method for marking data tags in an embodiment of the present application Figure 1 ;

[0046] Figure 6b A flowchart of a method for marking data tags in an embodiment of the present application Figure 2 ;

[0047] Figure 6c A flowchart of a method for marking data tags in an embodiment of the present application Figure 3 ;

[0048] Figure 6d A flowchart of a method for marking data tags in an embodiment of the present application Figure 4 ;

[0049] Figure 7 A schematic diagram of a data weighted graph provided in an embodiment of the present application;

[0050] Figure 8 A schematic diagram of a quantum architecture provided in an embodiment of the present application Figure 2 ;

[0051] Figure 9 A schematic diagram of a quantum architecture provided in an embodiment of the present application Figure 3 ;

[0052] Figure 10 A schematic diagram of a quantum architecture provided in an embodiment of the present application Figure 4 ;

[0053] Figure 11 A schematic diagram of a label weighted graph provided in an embodiment of the present application;

[0054] Figure 12A flowchart of a method for marking data tags in an embodiment of the present application Figure 5 ;

[0055] Figure 13 A schematic diagram of a quantum architecture provided in an embodiment of the present application Figure 5 ;

[0056] Figure 14 A schematic diagram of the hardware structure of a computing device provided in an embodiment of the present application;

[0057] Figure 15 A schematic diagram of the hardware structure of a quantum annealing device provided in an embodiment of the present application;

[0058] Figure 16 A schematic diagram of the structure of a chip system provided in an embodiment of the present application;

[0059] Figure 17 A schematic diagram of the structure of a computer program product provided in an embodiment of the present application. DETAILED DESCRIPTION

[0060] When performing data analysis, you can improve efficiency by labeling the data. Data can be any data from which features can be extracted, such as images or biometric data. Data labels are used to characterize or categorize the features of the data. Each piece of data has a label. Labeled data is called labeled data, while unlabeled data is called unlabeled data.

[0061] The embodiment of the present application provides a method for marking data tags, which can be applied to a system for marking data tags. The system includes a computing device and a quantum annealing device. The computing device determines a target control parameter based on at least one labeled data and at least one unlabeled data, and inputs the target control parameter into the quantum annealing device to obtain the state of the quantum bit (referred to as quantum in the embodiment of the present application) corresponding to the unlabeled data. Among them, the label of the data can be characterized by the state of each quantum in the quantum group corresponding to the data. The computing device decodes the quantum state output by the quantum annealing device to obtain the label of the unlabeled data. Compared with the prior art, the method for marking data tags provided by the embodiment of the present application can mark multiple tags at the same time, and can mark multiple unlabeled data at the same time, effectively improving the efficiency of data marking tags and reducing the demand for the amount of labeled data.

[0062] refer to Figure 1 , Figure 1 A data tagging system 10 is shown. The system 10 includes a computing device 20 and a quantum annealing device 30 connected to the computing device 20.

[0063] The computing device 20 may be a general-purpose computer, a classic computer, a personal computer (PC), a personal digital assistant (PDA), a netbook, a server, or any other computing device capable of implementing the embodiments of the present application, and the embodiments of the present application do not limit this.

[0064] refer to Figure 2 , Figure 2 FIG. 2 shows a hardware structure of a computing device 20 provided in an embodiment of the present application. Figure 2 As shown, the computing device 20 includes a processor 21, a memory 22, a communication interface 23, and a bus 24. The processor 21, the memory 22, and the communication interface 23 may be connected via the bus 24.

[0065] The processor 21 is the control center of the computing device 20 and can be a general-purpose central processing unit (CPU) or other general-purpose processors, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0066] As an example, the processor 21 may include one or more CPUs, such as Figure 2 CPU 0 and CPU1 are shown in Figure 1.

[0067] The memory 22 may be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM) or other type of dynamic storage device that can store information and instructions, an electrically erasable programmable read-only memory (EEPROM), a disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0068] In one possible implementation, the memory 22 may exist independently of the processor 21. The memory 22 may be connected to the processor 21 via a bus 24 and used to store data, instructions, or program code. When the processor 21 calls and executes the instructions or program code stored in the memory 22, the method for labeling data tags provided in the embodiments of the present application can be implemented.

[0069] In another possible implementation, the memory 22 may also be integrated with the processor 21 .

[0070] Communication interface 23 is used to connect computing device 20 to other devices (such as quantum annealing device 30) via a communication network. The communication network can be Ethernet, a radio access network (RAN), a wireless local area network (WLAN), etc. Communication interface 23 can include a receiving unit for receiving data and a sending unit for sending data.

[0071] The bus 24 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, Figure 2 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0072] It should be pointed out that Figure 2 The structure shown in the figure does not constitute a limitation on the computing device 20, except Figure 2 In addition to the components shown, the computing device 20 may include more or fewer components than shown, or combine certain components, or arrange the components differently.

[0073] The quantum annealing device 30 may be a quantum annealing machine, a quantum computer, or any other device capable of implementing the embodiments of the present application, which is not limited in the embodiments of the present application.

[0074] refer to Figure 3 , Figure 3 FIG. 1 shows a hardware structure diagram of the quantum annealing device 30 provided in an embodiment of the present application. Figure 3 As shown, Figure 3 The system includes a quantum processor 31, a communication interface 33 and a bus 34. The quantum processor 31 and the communication interface 33 can be connected via the bus 34.

[0075] The quantum processor 31 is the control center of the quantum annealing device 30 , and includes a quantum unit 311 and a control circuit unit 312 .

[0076] The quantum unit 311 includes a fixed number of quanta, which can form different quantum architectures.

[0077] In the quantum architecture of the quantum annealing device 30, there may be s quantum groups, each of which includes n quanta, where s and n are both positive integers greater than or equal to 1. As an example, if a physical quantity (such as an electron or photon) has a smallest indivisible basic unit that conforms to the basic assumptions of quantum mechanics and has quantum effects, then this physical quantity is quantized, and the smallest basic unit is called a quantum. In the quantum annealing device 30, the number of quanta is usually fixed, and this number is determined by the hardware structure of the quantum annealing device 30. Each quantum in the quantum annealing device 30 has a unique identity (ID).

[0078] Within a quantum group, different quanta can be uncorrelated. The state of each quantum is characterized by its eigenstate, which can be controlled by the longitudinal field applied to it. Measuring a physical quantity in a system will yield different results with varying probabilities. This probability can be determined by performing multiple measurements, and each measurement result belongs to an eigenstate of the measured quantity.

[0079] For example, if an electron is regarded as a quantum, the state of the quantum can be characterized by the spin direction of the electron. The spin direction of the electron can be upward or downward, that is, the state of the quantum can be upward or downward. Figure 4 As shown, the state of quantum 41 is upward, and the state of quantum 42 is downward.

[0080] Between quantum groups, the coupling correlation between the quantum groups in which the two quanta are located can be achieved by adjusting the size of the coupling effect between the two quanta.

[0081] refer to Figure 5 , Figure 5 FIG. 1 shows a quantum architecture diagram of a quantum annealing device 30 provided in an embodiment of the present application. Figure 5 As shown, Figure 5 The quantum group 51 includes quantum group 52, quantum group 53, quantum group 54, quantum group 55, quantum group 56, quantum group 57, and quantum group 58. Each quantum group includes two quantums. Figure 5 The quantum groups shown are fully interconnected, that is, there is a coupling effect between every two quantum groups. Figure 5 Indicated by black line.

[0082] Taking quantum group 51 as an example, there is no correlation between the two quanta in quantum group 51, that is, there is no coupling between the two quanta in quantum group 51. Quantum group 51 is coupled with quantum groups 52, 53, 54, 55, 56, 57, and 58 respectively.

[0083] certainly, Figure 5 The quantum architecture shown is for illustrative purposes only. In practical applications, coupling can also exist between quantum quanta within a quantum group, and the coupling between two quantum groups in a quantum annealing architecture can also be non-fully interconnected, which is not limited in this embodiment of the application.

[0084] The control circuit unit 312 is used to apply a longitudinal field to the quantum in the quantum architecture to control the state of the quantum. The control circuit unit 312 is also used to control and adjust the magnitude of the coupling between quantum groups in the quantum architecture.

[0085] The communication interface 33 may refer to the description of the communication interface 23 in the above-mentioned computing device 20 , and the bus 34 may refer to the description of the bus 24 in the above-mentioned computing device 20 , which will not be repeated here.

[0086] The following is a brief description of some of the terms and technologies involved in the embodiments of this application:

[0087] 1) Hamiltonian

[0088] Hamiltonian is used to describe the total energy of the system. In the quantum annealing device 30, Hamiltonian H can be expressed by the following formula (1):

[0089] Formula (1):

[0090] Among them, A(s) and B(s) are functions of time, h i represents the magnitude of the longitudinal field applied to the i-th quantum. i，j It represents the coupling effect between the i-th and j-th quanta. represents the state of the i-th quantum, In the above formula (1), the first term on the right side of the equal sign represents the initial state energy of quantum annealing, and the second term on the right side of the equal sign represents the final state energy at the end of quantum annealing.

[0091] It should be noted that when h in the above formula (1) i The larger the value, the greater the probability that the state of the i-th quantum remains unchanged during the quantum annealing process. i,j The larger the value, the more similar the states of quantum i and quantum j are. i,jWhen it is large enough, the states of the i-th quantum and the j-th quantum are both up, or both down.

[0092] 2) Data weighted graph

[0093] A data-weighted graph consists of multiple nodes, each representing a piece of data. Edges between nodes represent the similarity between the two pieces of data represented by the two nodes connected by the edge. The weight of an edge is related to the degree of similarity represented by the edge; the greater the similarity, the greater the weight of the edge.

[0094] Exemplarily, a data weighted graph may include a first node and a second node, where the first node represents data 1, the second node represents data 2, and the edge between the first node and the second node represents the similarity between data 1 and data 2. The greater the similarity between data 1 and data 2, the greater the weight of the edge between the first node and the second node.

[0095] The first node and the second node may be any two nodes in the data weighted graph.

[0096] The weight of the above edge can be expressed by a numerical value or by the length of the edge.

[0097] 3) Other terms

[0098] In the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0099] In the embodiments of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0100] The following describes the method for marking data labels provided in the embodiments of the present application in conjunction with the accompanying drawings.

[0101] Example 1

[0102] Figure 6a The flowchart of the method for marking data tags provided by the embodiment of the present application is shown. The method can be applied to Figure 1 The system 10 shown includes the following steps:

[0103] S601. The computing device 20 obtains s data.

[0104] The s data include x labeled data and y unlabeled data. s = x + y, where x and y are integers greater than or equal to 1, and s is an integer greater than or equal to 2. The x labeled data have m types of labels, where m is an integer greater than or equal to 1.

[0105] The computing device 20 may obtain the s data by downloading from the network, or by receiving user input or sending from other devices, which is not limited in the present embodiment.

[0106] S602: The computing device 20 determines the number of encoding bits for encoding the tags in the tagged data according to the number of tag types in the tagged data.

[0107] Optionally, a coding bit has two states, for example, "0" and "1".

[0108] In one implementation, if computing device 20 encodes the labels in the labeled data using a one-hot encoding scheme, then m encoding bits can represent m types of labels. That is, computing device 20 determines the number of encoding bits to be used to encode the labels in the labeled data as m based on the number m of label types in the labeled data.

[0109] In another implementation, the computing device 20 uses a binary encoding method to encode the label in the labeled data, and n encoding bits can represent 2 n When the number of types of tags in the tag data is m, the computing device 20 determines that the number of encoding bits n is greater than or equal to: the result of log2m is rounded up, where n is a positive integer greater than or equal to 1.

[0110] For example, if the tag data includes four tags, the computing device 20 determines that the number of encoding bits is log24, that is, the value of n is 2. If the tag data includes three tags, the computing device 20 determines that the number of encoding bits required is log23 rounded up, that is, the value of n is 2.

[0111] S603: The computing device 20 encodes the tag with the tagged data according to the determined number of encoding bits.

[0112] The computing device 20 may encode the tag with the tagged data using the following method 1 or method 2.

[0113] It can be understood that, in labeled data, data with the same label have the same corresponding encoding results.

[0114] Method 1: The computing device 20 uses a one-hot encoding method to encode the labels of the labeled data.

[0115] When the computing device 20 encodes the labels in the labeled data using a one-hot encoding method, in S602 above, the computing device 20 determines that m types of labels are encoded using m encoding bits, that is, the m types of labels are represented by the states of m encoding bits.

[0116] For example, if the label data includes three labels (including label 1, label 2, and label 3), the number of encoding bits required for one-hot encoding is 3. Computing device 20 encodes the three labels using one-hot encoding as follows: label 1 is encoded as "001", label 2 is encoded as "010", and label 3 is encoded as "100". The encodings of the three labels differ by two bits.

[0117] Method 2: The computing device 20 uses a binary encoding method to encode the tag with the tag data.

[0118] For example, if the tag data includes three tags (including tag 1, tag 2, and tag 3), and the number of encoding bits is 2, the computing device 20 encodes the three tags using binary encoding: tag 1 is encoded as "00", tag 2 is encoded as "01", and tag 3 is encoded as "10".

[0119] S604: The computing device 20 determines the similarity between every two data in the s data to obtain a data weighted graph.

[0120] The data weighted graph is used to represent the similarity between s data and every two data in the s data.

[0121] Specifically, computing device 20 may use a feature extraction algorithm to extract features from each of the s data. For example, if data 1 and data 2 are both image data among the s data, computing device 20 may use a feature extraction algorithm to reduce the dimensions of the pixels of data 1 and data 2 to a two-dimensional plane and extract features from data 1 and data 2.

[0122] The computing device 20 then uses a similarity algorithm based on the characteristics of each data item to calculate the similarity between each pair of data items. The similarity algorithm may include a cosine similarity algorithm or a Euclidean distance algorithm. The specific process of calculating the similarity between data items using the similarity algorithm employed by the computing device 20 can be found in the prior art and will not be further described here.

[0123] Optionally, the computing device 20 may use a numerical value between 0 and 1 to represent the similarity.

[0124] After the computing device 20 determines the similarity between each two data in the s data, a data weighted graph of the s data is obtained. At this time, each two data in the s data can be associated by the similarity between the two data.

[0125] In a data weighted graph, each node represents a piece of data, and each edge represents the similarity between the two pieces of data connected by the edge. For example, if the value of s is 6, refer to Figure 7 , Figure 7 The data weighted graph 70 shown in FIG is used to represent data 1-6 and the similarity between each pair of data 1-6. Specifically, data weighted graph 70 includes six nodes: node 71 represents data 1, node 72 represents data 2, node 73 represents data 3, node 74 represents data 4, node 75 represents data 5, and node 76 represents data 6. The edge between each pair of nodes represents the similarity between the data represented by the two nodes. For example, edge 77 represents the similarity between data 5 and data 6.

[0126] In the embodiment of the present application, the nodes in the data weighted graph have the same concept as the data in the data weighted graph, and the two can be used interchangeably.

[0127] The embodiment of the present application does not limit the execution order between S602-S603 and S604. For example, S602-S603 can be executed simultaneously with S604, or S602-S603 can be executed first and then S604.

[0128] S605 : The computing device 20 determines the target control parameters according to the data weighted graph.

[0129] The target control parameters include: a longitudinal field applied to each quantum in a quantum group corresponding to labeled data, and a coupling effect applied between a quantum group corresponding to each unlabeled data and other quantum groups except the quantum group.

[0130] When the data weighted graph includes a labeled data (corresponding to the first labeled data in the embodiment of the present application) and an unlabeled data (corresponding to the first unlabeled data in the embodiment of the present application), the above-mentioned target control parameters include: the longitudinal field applied to the quantum in the quantum group corresponding to the first labeled data, and the coupling effect applied between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data.

[0131] The specific process of the computing device 20 determining the target control parameter according to the data weighted graph can be referred to the description below and will not be repeated here.

[0132] S606 : The computing device 20 sends target control parameters to the quantum annealing device 30 .

[0133] The computing device 20 may send the target control parameters to the quantum annealing device 30 through the communication interface 23. Specifically, the computing device 20 may send the target control parameters to the quantum annealing device 30 through the following method 1 or method 2.

[0134] In a first approach, the computing device 20 may send the target control parameters determined in S605 to the quantum annealing device 30 .

[0135] Method 2: If the encoding result of a tag is represented by N bits, the computing device 20 can send the longitudinal field of the quantum corresponding to the first encoding bit of each data tag and the first control parameter of the coupling between the quantum corresponding to the first encoding bit of each pair of data tags to the quantum annealing device 30 to perform quantum annealing to obtain the first annealing result. Next, the computing device 20 sends the longitudinal field of the quantum corresponding to the second encoding bit of each data tag and the second control parameter of the coupling between the quantum corresponding to the second encoding bit of each pair of data tags to the quantum annealing device 30 to perform quantum annealing to obtain the second annealing result. Similarly, the computing device 20 sends the longitudinal field of the quantum corresponding to the nth encoding bit of each data tag and the nth control parameter of the coupling between the quantum corresponding to the nth encoding bit of each pair of data tags to the quantum annealing device 30 to perform quantum annealing to obtain the nth annealing result. Where N is an integer greater than or equal to 1, 1≤n≤N, and n is an integer.

[0136] As can be seen from the above description, the target control parameters in the above-mentioned method 1 include the first control parameter, the second control parameter, ..., and the nth control parameter in the above-mentioned method 2. In response to the operation of the computing device 20, the quantum annealing device 30 receives the first control parameter, the second control parameter, ..., and the nth control parameter through the communication interface 33 and performs n quantum annealing, thereby obtaining n annealing results. The first annealing result to the nth annealing result are collectively used to represent the label of the unlabeled data.

[0137] It can be seen that through the above-mentioned second approach, the quantum resources required by the quantum annealing device 30 during quantum annealing are reduced, thereby effectively reducing the hardware implementation complexity of the quantum annealing device 30 during quantum annealing.

[0138] S607 , the quantum annealing device 30 performs quantum annealing according to the target control parameters, and after the quantum annealing is completed, sends the current state of the quantum in the quantum group corresponding to the unlabeled data to the computing device 20 .

[0139] Optionally, after receiving the target control parameters, the quantum annealing device 30 may perform quantum annealing according to the default annealing starting temperature, annealing final temperature, and annealing speed of the quantum annealing device 30 .

[0140] Optionally, after receiving the target control parameters, the quantum annealing device 30 can allow the user to set the annealing start temperature, annealing end temperature, and annealing speed through the setting interface of the quantum annealing device 30. The quantum annealing device 30 performs quantum annealing according to the target control parameters, annealing start temperature, annealing end temperature, and annealing speed.

[0141] Optionally, the user can first set the annealing start temperature, annealing final temperature and annealing speed through the setting interface of the quantum annealing device 30, and after the quantum annealing device 30 receives the target control parameters, perform quantum annealing according to the target control parameters, annealing start temperature, annealing final temperature and annealing speed.

[0142] The above-mentioned annealing starting temperature, annealing final temperature and annealing speed can be set according to actual conditions, and the embodiments of the present application do not specifically limit them.

[0143] Specifically, the above-mentioned annealing start temperature can be the highest starting temperature supported by the hardware of the quantum annealing device 30, and the above-mentioned annealing final temperature can be the lowest final temperature supported by the hardware of the quantum annealing device 30. Of course, the above-mentioned annealing start temperature and annealing final temperature can also be any temperature between the highest starting temperature and the lowest final temperature supported by the hardware of the quantum annealing device 30, and this embodiment of the application is not limited to this. Among them, the annealing start temperature is higher than the annealing final temperature.

[0144] After the quantum annealing is completed, the quantum annealing device 30 sends the states of the quanta in the quantum group without the longitudinal field to the computing device 20 through the communication interface 33 , that is, only sends the states of the quanta in the quantum group corresponding to the unlabeled data to the computing device 20 .

[0145] In response to the operation of the quantum annealing device 30 , the computing device 20 receives the states of the quanta in the quantum group corresponding to the unlabeled data through the communication interface 23 .

[0146] S608. The computing device 20 decodes the received unlabeled data according to the quantum state corresponding to the unlabeled data to obtain the label of the unlabeled data.

[0147] Corresponding to the above S607, the computing device 20 sends the target control parameters to the quantum annealing device in different ways. The two cases are described below.

[0148] Method 1: The computing device 20 determines the binary code or one-hot code corresponding to the quantum state according to the state of the quantum in the quantum group corresponding to the received unlabeled data and the second preset rule.

[0149] The computing device 20 decodes the determined binary code or one-hot code to obtain a label corresponding to the binary code or one-hot code, that is, obtains a label for the unlabeled data.

[0150] Method 2: The computing device 20 determines the binary code or one-hot code corresponding to the quantum state according to the states of the n quanta received n times and the second preset rule.

[0151] Exemplary, reference Figure 8 , if the state of the quantum of quantum group 85 received by the computing device 20 for the first time is "up", and the state of the quantum of quantum group 85 received for the second time is "down", according to the second preset rule, the computing device 20 determines that the binary code corresponding to the quantum group 85 is "01".

[0152] The computing device 20 decodes the determined binary code or one-hot code to obtain a label corresponding to the binary code or one-hot code, that is, obtains a label for the unlabeled data.

[0153] It is easy to understand that the above decoding corresponds to the encoding in S603, that is, the computing device 20 decodes the determined binary encoding or one-hot encoding according to the encoding rule in S603. This will not be described in detail here.

[0154] In practical applications, the binary encoding result obtained by computing device 20 based on the quantum state output by quantum annealing device 30 may not correspond to a pre-encoded label. For example, when computing device 20 uses binary encoding to encode three types of tagged data, label 1 is encoded as "00", label 2 is encoded as "01", and label 3 is encoded as "10". However, there is no corresponding pre-encoded label for the encoding combination "11". The following scenarios illustrate this situation.

[0155] Scenario 1: The quantum annealing device 20 assigns a quantum group to each data in the data weighted graph. The similarity between each two data corresponds to the coupling effect between the quantum groups corresponding to the two data. For a detailed description of this scenario, please refer to the description of S6051A to S6054A below.

[0156] In this scenario, if after quantum annealing is performed, if the binary coding result obtained by the computing device 20 according to the state of the quantum output by the quantum annealing device 30 according to the annealing result does not correspond to a coded label, a label is assigned to the binary coding result according to the fourth preset rule.

[0157] The fourth preset rule may include: the computing device 20 randomly assigning a label to the binary encoding result. For example, when the computing device 20 encodes three types of labels of the tagged data using a binary encoding method, label 1 is encoded as "00", label 2 is encoded as "01", and label 3 is encoded as "10", when the binary encoding result corresponding to the annealing result is "11", the encoding result "11" may be randomly assigned label 1, label 2, or label 3.

[0158] Alternatively, the fourth preset rule may include: the computing device 20 assigns a label to the binary coding result in the annealing result that has no corresponding encoded label based on the probability of the encoded label corresponding to the binary coding result corresponding to each annealing result in multiple annealing results and a random number.

[0159] The random number may be randomly assigned to the data corresponding to each quantum group in each annealing result when the computing device 20 decodes the annealing result after receiving it. Alternatively, the random number may be randomly assigned to the annealing result that does not correspond to an encoded tag when the computing device 20 decodes the annealing result after receiving it. Alternatively, the random number may be randomly assigned to the data corresponding to each quantum group in each annealing result when the quantum annealing device 30 sends the annealing result to the computing device 20. The random number may be a value between 0 and 99.

[0160] Exemplarily, when the computing device 20 encodes three types of labels with tagged data using a binary encoding method, label 1 is encoded as "00", label 2 is encoded as "01", and label 3 is encoded as "10". After 10 quantum annealings, for a quantum group corresponding to a data, the binary encoding results obtained from the three annealing results do not correspond to an encoded label, and the random number in the first annealing result is 8, the random number in the second annealing result is 26, and the random number in the third annealing result is 55. The binary encoding results obtained from the other seven annealing results correspond to different encoded labels. For example, among the seven annealing results, the binary encoding result obtained from the first annealing result corresponds to label 1, the binary encoding results obtained from the second annealing result all correspond to label 2, and the binary encoding results obtained from the fourth annealing result all correspond to label 3.

[0161] In this case, among the seven annealing results, the probability that the annealing result corresponds to label 1 is 1 / 7, or 14%, and the corresponding random number range is 0-13. The probability that the annealing result corresponds to label 2 is 2 / 7, or 29%, and the corresponding random number range is 14-42. The probability that the annealing result corresponds to label 3 is 4 / 7, or 57%, and the corresponding random number range is 43-99. In this case, the random number 8 from the first annealing result falls within the range of 0-13, that is, the probability range (14%) for the annealing result to correspond to label 1. Therefore, the computing device 20 assigns label 1 to the binary encoding result corresponding to the first annealing result. The random number 26 from the second annealing result falls within the range of 14-42, that is, the probability range (29%) for the annealing result to correspond to label 2. Therefore, the computing device 20 assigns label 2 to the binary encoding result corresponding to the second annealing result. The third random number 55 falls within the range of 43 to 99, that is, the random number 55 falls within the probability range (57%) of the annealing result corresponding to label 3. In this case, the computing device 20 assigns label 3 to the binary coding result corresponding to the third annealing result.

[0162] Alternatively, the fourth preset rule may include: the computing device 20 assigning the label corresponding to the encoding result closest to the binary encoding result to the binary encoding result. For example, when the computing device 20 encodes three types of labels of the labeled data using a binary encoding method, label 1 is encoded as "00", label 2 is encoded as "01", and label 3 is encoded as "10". When the binary encoding result corresponding to the annealing result is "11", since the encoding results "01" and "10" both differ from the encoding result "11" by one bit, the computing device 20 may assign the label corresponding to the encoding result "01" or "10" to the encoding result "11".

[0163] Scenario 2: The quantum annealing device assigns a quantum group to each piece of data in the weighted graph. Coupling exists between these quantum groups. For a detailed description of this scenario, see S6051B to S6054B below.

[0164] In this scenario, if after executing quantum annealing, the binary encoding result obtained by computing device 20 based on the quantum states output by quantum annealing device 30 does not correspond to a coded label, computing device 20 can increase the number of data with a connection relationship so that the binary encoding result obtained from the quantum states output by quantum annealing device 30 can all correspond to a coded label. If, after increasing the number of data with a connection relationship, the binary encoding result obtained from the quantum states output by quantum annealing device 30 still does not correspond to a coded label, computing device 20 can assign a label to the binary encoding result according to the fourth preset rule described above. The increase in the number of data with a connection relationship can be referred to in the description of S6051B below and will not be further elaborated here.

[0165] Next, another encoding method provided in the embodiment of the present application when binary encoding is performed on the tag with the tagged data in S603 is described.

[0166] When a large number of different types of labels are present in the data, i.e., when the value of m is large, the computing device 20 may randomly perform binary encoding for each of the m types of labels when performing step S603 to encode the labels of the labeled data. Of course, the computing device may also choose to perform binary encoding for each of the m types of labels using the following method to improve the accuracy of the data labeling method provided in the embodiment of the present application.

[0167] In the embodiment of the present application, the method is described by taking m as 4 (including label 1, label 2, label 3, and label 4), and the number of labeled data corresponding to labels 1 to 4 are x1, x2, x3, and x4 respectively.

[0168] Optional, reference Figure 6b , S603 further includes the following steps:

[0169] S6031. The computing device 20 extracts features of all labeled data corresponding to each of the four labels, and calculates the average value of the features of all labeled data corresponding to each label.

[0170] Specifically, the computing device 20 may refer to the description of extracting features of data in the above S604 to extract features of all labeled data corresponding to each of the four labels, which will not be described in detail in this embodiment of the present application.

[0171] For example, the computing device 20 extracts x1 features of the labeled data corresponding to label 1 and calculates the average value a1 of the x1 features. Similarly, the computing device 20 calculates the average values ​​a2, a3, and a4 of the features corresponding to labels 2 to 4.

[0172] S6032: The computing device 20 calculates the similarity between every two tags based on the average value of the features corresponding to each tag to obtain a tag weighted graph.

[0173] Specifically, the computing device 20 may refer to the description of calculating the similarity between data in S604 above to calculate the similarity between every two of the four tags, which will not be described in detail in this embodiment of the present application.

[0174] For example, referring to Table 1, Table 1 shows that the computing device 20 calculates the similarity between each two tags among the four tags based on the average value of the features corresponding to each tag:

[0175] Table 1

[0176] Label Similarity Label 1 and Label 2 <![CDATA[s 1-2 ]]> Label 1 and Label 3 <![CDATA[s 1-3 ]]> Label 1 and Label 4 <![CDATA[s 1-4 ]]> Label 2 and Label 3 <![CDATA[s 2-3 ]]> Label 2 and Label 4 <![CDATA[s 2-6 ]]> Label 3 and Label 4 <![CDATA[s 3-4 ]]>

[0177] The above four labels and the similarity between each two labels constitute a label weighted graph.

[0178] Each node in the label-weighted graph represents a label, and an edge in the label-weighted graph represents the similarity between the two labels corresponding to the two nodes connected by the edge. The greater the similarity between the two labels, the greater the weight of the edge in the label-weighted graph corresponding to that similarity, and the shorter the edge.

[0179] Exemplary, reference Figure 11 , Figure 11 The label weighted graph 110 is shown, which is composed of the above four labels and the similarity between each two labels. Each node in the label weighted graph 110 represents a label, and the edge of the label weighted graph 110 corresponds to the similarity between the two labels corresponding to the two nodes connected by the edge. For example, the edge 114 between label 1 and label 4 represents the similarity s 1-4 , the weight of edge 114 and the similarity s 1-4 correspond.

[0180] S6033. The computing device 20 determines a first label and a second label among the four labels of the data, and determines a first path (corresponding to the first sequence in the embodiment of the present application) based on the first label and the second label.

[0181] Among them, the similarity between the first label and the second label is the greatest.

[0182] Specifically, the computing device 20 determines the first tag and the second tag corresponding to the maximum similarity value based on the similarity between each two tags in the four tags determined in S6032. For example, the similarity between tag 1 and tag 4 is s 1-4The computing device 20 determines that tag 1 is the first tag and tag 4 is the second tag. Alternatively, the computing device 20 determines that tag 2 is the first tag and tag 1 is the second tag. The computing device 20 traverses the tag weighted graph according to the positions of the first tag and the second tag in the tag weighted graph to determine a first path in the tag weighted graph.

[0183] The starting point and end point of the first path are the first label and the second label determined above, respectively. The first path passes through each node in the label-weighted graph once, and the weight of the first path is the maximum weight among all paths that pass through each node in the weighted graph once. The weight of a path is the sum of the weights of each edge in the label-weighted graph that the path passes through.

[0184] Exemplary, reference Figure 11 , if the first label is label 1 and the second label is label 4. Then, in label-weighted graph 110, path 1, which passes through each node once, is: label 1 → label 2 → label 3 → label 4. The weight A of path 1 is the sum of the weights of edges 111, 112, and 113. Path 2, which passes through each node once, is: label 1 → label 3 → label 2 → label 4. The weight A of path 2 is the sum of the weights of edges 115, 112, and 116. If A is greater than B, computing device 20 determines that path 1 is the first path.

[0185] When the number of nodes in the label-weighted graph is large, the computing device 20 can determine the first path by traversing each node in the label-weighted graph. The computing device 20 can also determine the first path in the label-weighted graph by mapping the label-weighted graph to a quantum annealing device using a quantum annealing method. This embodiment of the present application is not limited to this.

[0186] S6034. The computing device 20 performs Gray coding on the label corresponding to each node in sequence according to the order of the nodes in the first path.

[0187] After determining the first path, the computing device 20 sequentially performs Gray coding on the labels corresponding to the nodes on the first path according to the order of the nodes in the first path.

[0188] For example, if the first label is label 1 and the second label is label 4, and the first path is: label 1 → label 2 → label 3 → label 4, then the computing device 20 starts from label 1 and encodes each label in sequence using Gray coding according to the order of the midpoints in the first path. Refer to Table 2, which shows the results of computing device 20 encoding each label using Gray coding according to the order of the first path:

[0189] Table 2

[0190] Label Gray coding Tag 1 00 Tag 2 01 Tag 3 11 Tag 4 10

[0191] It can be seen that after Gray coding the labels represented by each node in the first path, the difference between two adjacent codes is only one bit. In other words, the codes corresponding to two labels with a small similarity in the label weighted graph only differ by one bit.

[0192] Thus, the data labeling method provided in the embodiments of the present application uses quantum annealing to perform semi-supervised learning based on labeled and unlabeled data. During quantum annealing, corresponding quantum groups are assigned to both labeled and unlabeled data, thereby simultaneously assigning m types of labels to multiple unlabeled data, improving the efficiency of data labeling. Here, semi-supervised learning refers to learning using both labeled and unlabeled data.

[0193] The following describes the process of the computing device 20 determining the target control parameters in the above S605 in different scenarios.

[0194] Scenario 1: The quantum annealing device 20 assigns a quantum group to each data in the data weighted graph, and the similarity between each two data corresponds to the coupling effect between the quantum groups corresponding to the two data.

[0195] Based on scenario 1, reference Figure 6c The computing device 20 may determine the target control parameters by the following steps:

[0196] S6051A: The computing device 20 assigns a quantum ID to each data in the data weighted graph. The quantum identified by the quantum ID constitutes a quantum group corresponding to the data.

[0197] Specifically, the computing device 20 assigns a quantum ID to each data from pre-stored quantum IDs according to the data weighted graph and the first preset rule, so as to obtain s quantum groups corresponding to the s data.

[0198] The quantum IDs pre-stored in computing device 20 include the IDs corresponding to the quanta in quantum annealing device 30. Each quantum ID uniquely identifies a quantum in quantum annealing device 30. This embodiment of the present application does not limit the content of the quantum ID; for example, the quantum ID may be a unique number corresponding to each quantum.

[0199] The first preset rule may be that the computing device 20 randomly assigns a quantum ID to each data, or may be that the quantum ID is assigned to each data according to a certain rule, which is not limited in this embodiment of the present application.

[0200] It should be noted that the number of quanta corresponding to the quantum IDs assigned by the computing device 20 to the s data in the data weighted graph is less than or equal to the total number of quanta included in the quantum annealing device 30 .

[0201] In one implementation, computing device 20 assigns n quantum IDs to each data item in the data weighted graph, where n can be the number of encoding bits determined in S602. In other words, the number of quantum IDs assigned by computing device 20 to each data item is the same as the number of encoding bits determined in S602. This method corresponds to the method described in S606.

[0202] In another implementation, the computing device 20 assigns a quantum ID to the first bit of the encoding result of the label of each data item in the data-weighted graph for the first time, thereby forming a quantum group corresponding to the data item, and then performs a subsequent step to determine the first control parameter. The computing device 20 assigns a quantum ID to the second bit of the encoding result of the label of each data item in the data-weighted graph for the second time, thereby forming a quantum group corresponding to the data item, and then performs a subsequent step to determine the second control parameter. Similarly, the computing device 20 assigns a quantum ID to the nth bit of the encoding result of the label of each data item in the data-weighted graph for the nth time, thereby forming a quantum group corresponding to the data item, and then performs a subsequent step to determine the nth control parameter.

[0203] As can be seen, the target control parameters in the first implementation include the first, second, ..., and nth control parameters in the second implementation. In other words, computing device 20 uses time division multiplexing to assign a quantum ID to the corresponding encoded bit of each data tag at a time and then performs subsequent steps to determine the corresponding control parameters. This method of assigning quantum IDs corresponds to the second method in S606 described above.

[0204] It is understood that when the method for tagging data provided in the embodiments of the present application begins execution, computing device 20 has already determined whether to send the target control parameters using method 1 or method 2 in S606 to perform quantum annealing and obtain the annealing results. Accordingly, when assigning a quantum ID to each data item, computing device 20 can use the implementation method corresponding to the determined method for sending the target control parameters, of the two aforementioned implementation methods, to assign a quantum ID to each data item.

[0205] For example, if s is 6, and among the 6 data, the labeled data are data 1, data 2, data 3, and data 4, and include 4 types of labels, and the unlabeled data are data 5 and data 6. The computing device 20 determines that the number of coding bits for the 4 types of labels is 2, and also determines to allocate the same number of quanta as the number of label coding bits to each data, that is, 2 quanta. Figure 8 , Figure 8 It shows that the computing device 20 allocates two quanta to each of the six data, and these two quanta constitute the quantum group corresponding to each data. Figure 8 As shown, data 1 corresponds to quantum group 81, data 2 corresponds to quantum group 82, data 3 corresponds to quantum group 83, data 4 corresponds to quantum group 84, data 5 corresponds to quantum group 85, and data 6 corresponds to quantum group 86.

[0206] The data in the data weighted graph can be represented by quantum groups in the quantum annealing device 30 , one data corresponds to one quantum group, and the similarity between the data can be represented by the coupling between the quantum groups corresponding to the data.

[0207] S6052A: The computing device 20 determines the magnitude of the coupling effect between the quantum group corresponding to each unlabeled data and other quantum groups except the quantum group.

[0208] The computing device 20 determines the coupling effect between the quantum group corresponding to the unlabeled data and the other quantum groups except the quantum group according to the similarity between each unlabeled data and the other data except the data in the data weighted graph and the first preset relationship.

[0209] Here, the first preset relationship is a preset relationship pre-stored by the computing device 20. The first preset relationship is used to record the corresponding relationship between the similarity and the coupling effect. The first preset relationship can be represented in a table or any other form, and this embodiment of the application is not limited to this.

[0210] Exemplary, reference Figure 8 , Figure 8 As shown by the black lines in the figure, the quantum group 85 corresponding to the unlabeled data and the quantum groups other than quantum group 85 (quantum group 81, quantum group 82, quantum group 83, quantum group 84, quantum group 86) are all coupled and connected, and the quantum group 86 corresponding to the unlabeled data and the quantum groups other than quantum group 86 are all coupled and connected.

[0211] The coupling between two quantum groups is essentially the coupling between the quanta of corresponding bits in the two quantum groups. Here, corresponding bits refer to two quanta with the same bit position in different quantum groups. For example, the quanta in two quantum groups that belong to the same first bit position are quanta of corresponding bits.

[0212] Exemplary, reference Figure 9 , Figure 9Figure 2 shows how quantum groups 85 and 86 are coupled. Quantum group 85 includes quantum 851 and quantum 852, where quantum 851 is the first bit and quantum 852 is the second bit. Quantum group 86 includes quantum 861 and quantum 862, where quantum 861 is the first bit and quantum 862 is the second bit. Therefore, quantum 851 and quantum 861 are quanta of corresponding bits, and quantum 852 and quantum 862 are quanta of corresponding bits. When quantum groups 85 and 86 need to be coupled, this is achieved through coupling between quantum 851 and quantum 861, and between quantum 852 and quantum 862. Furthermore, the ratio of the coupling effect 561 between quantum 851 and quantum 861 to the coupling effect 562 between quantum 852 and quantum 862 is 1:1.

[0213] S6053A: The computing device 20 determines the longitudinal field applied to each quantum in the quantum group corresponding to the label data.

[0214] Specifically, computing device 20 determines the direction of the longitudinal field applied to each quantum in the quantum group corresponding to the labeled data, as well as the magnitude of the longitudinal field. The direction of the longitudinal field is used to control the state of the quantum, while the magnitude of the longitudinal field is used to maintain the state of the quantum during the annealing process.

[0215] The process includes the following steps:

[0216] Step 1: The computing device 20 determines the direction of the longitudinal field applied to each quantum according to the state of the quantum in the quantum group corresponding to the label data.

[0217] Specifically, the computing device 20 determines the state of each quantum in the quantum group corresponding to the labeled data based on the label encoding result of the labeled data and the second preset rule. The computing device 20 determines the longitudinal field direction applied to each quantum based on the determined state of the quantum.

[0218] The second preset rule includes: one state of the coding bit corresponds to the quantum state "up", and the other state of the coding bit corresponds to the quantum state "down". For example, the coding bit "0" corresponds to the quantum state "down", and the coding bit "1" corresponds to the quantum state "up", or the coding bit "0" corresponds to the quantum state "down", and the coding bit "1" corresponds to the quantum state "up". This embodiment of the application does not limit this.

[0219] The following description of the embodiment of the present application is based on the second preset rule that the coding bit "0" corresponds to the quantum state "up", and the coding bit "1" corresponds to the quantum state "down".

[0220] For example, Figure 8 As shown, Figure 8 The quanta in the quantum group corresponding to the label data shown in FIG are represented by hollow circles, and the state of each quantum in the quantum group corresponding to the label data is represented by an arrow. For example, the states of quantum 811 and quantum 812 in quantum group 81 are both upward. Figure 8 The quanta in the quantum group corresponding to the unlabeled data shown in are represented by solid circles, and the states of the quanta in the quantum group corresponding to the unlabeled data are unknown.

[0221] It can be seen that the state of each quantum in the quantum group represents the label of the data corresponding to the quantum group.

[0222] Step 2: After determining the direction of the longitudinal field applied to each quantum in the quantum group corresponding to the label data, the computing device 20 determines the magnitude of the longitudinal field.

[0223] For a quantum group corresponding to label data, the magnitude of the longitudinal field applied to each quantum in the quantum group by the computing device 20 may be a first preset value or the maximum longitudinal field value supported by the hardware of the quantum annealing device 30. The first preset value may be specified by the user and is not limited in this embodiment of the present application.

[0224] It should be noted that the value of the longitudinal field applied to each quantum in the quantum group corresponding to the label data needs to be large enough, that is, h in formula (1) i is large enough. At this time, the state of the quantum can remain unchanged during the annealing process, or the probability that the state of the quantum can remain unchanged during the annealing process is extremely high. i The specific method for determining the value of can refer to the existing technology and will not be repeated here.

[0225] Optionally, the target control parameter may also be used to indicate the coupling effect between two quanta with the same state in the quanta corresponding to the tag data, that is, the target control parameter may also include: the coupling effect between two quanta with the same state in the quanta corresponding to the tag data. In this case, S605 in scenario 1 may also include:

[0226] S6054A (optional): The computing device 20 determines the coupling effect between two quanta with the same state among the quanta corresponding to the tag data.

[0227] In practical applications, the longitudinal field applied to each quantum in the quantum group corresponding to the labeled data may not be sufficient to maintain the quantum's state unchanged during the annealing process. Therefore, the computing device 20 can apply the maximum coupling effect supported by the hardware of the quantum annealing device 30 between two quanta in the same state among the quanta corresponding to the labeled data. Alternatively, the computing device 20 can apply a coupling effect of a second preset value between two quanta in the same state among the quanta corresponding to the labeled data. Here, the second preset value can be specified by the user and is not limited in this embodiment of the application.

[0228] It should be noted that the coupling effect between two quanta with the same state in the quanta corresponding to the labeled data needs to be sufficiently large, that is, J(i, j) in formula (1) is sufficiently large. In this way, during the quantum annealing process, this coupling effect can make the states of the two quanta tend to be the same. In other words, by applying a sufficiently large coupling effect between two quanta with the same state in the quanta corresponding to the labeled data, the probability of the two quanta changing state during the quantum annealing process is reduced.

[0229] Scenario 2: The quantum annealing device assigns a quantum group to each piece of data in the weighted graph, where some of the quantum groups are coupled to each other.

[0230] Based on scenario 2, reference Figure 6d The computing device 20 may determine the control parameters by the following steps:

[0231] S6051B: The computing device 20 determines data in the data weighted graph that needs to be allocated to subgroups.

[0232] In one possible implementation, for each data in the data weighted graph, the computing device 20 sorts the similarity values ​​represented by the edges connected to the data, and determines the edges corresponding to similarity values ​​with sorted values ​​greater than or equal to k, and determines the data connected to the edges, where k is a positive integer greater than or equal to 1. In this embodiment of the application, the data determined by the computing device 20 based on the data is referred to as a candidate data set corresponding to the data. In the candidate data set of the data, the data has a connection relationship with each data in the candidate data set.

[0233] Furthermore, the computing device 20 determines that the data to be allocated to the subgroup includes: each unlabeled data and data having a connection relationship with the unlabeled data.

[0234] Exemplary, reference Figure 7, if the value of k is 2, data 1, data 2, data 3, and data 4 are labeled data, and data 5 and data 6 are unlabeled data. The computing device 20 sorts the five similarity values ​​represented by the five edges (edge ​​77, edge 78, edge 79, edge 710, and edge 711) connected to node 75 representing data 5, and determines the edges with the largest and second largest similarity values. For example, the computing device 20 determines that the similarity represented by edge 77 is the largest, and the similarity represented by edge 79 is the second largest. Then, the computing device 20 determines that the data represented by another node 76 connected to edge 77 connected to node 75 is data 6, and the data represented by another node 72 connected to edge 79 connected to node 75 is data 2. In other words, the computing device 20 can determine data 2 and data 6 from data 5, that is, data 2 and data 6 are respectively connected to data 5. Similarly, the computing device 20 can determine data 1 and data 2 from data 6, that is, data 1 and data 2 are respectively connected to data 6. The computing device 20 can determine data 2 and data 6 from data 1, that is, data 2 and data 6 are respectively connected to data 1. The computing device 20 can determine data 1 and data 6 from data 2, that is, data 1 and data 6 are respectively connected to data 2. The computing device 20 can determine data 2 and data 4 from data 3, that is, data 2 and data 4 are respectively connected to data 3. The computing device 20 can determine data 3 and data 5 from data 4, that is, data 3 and data 5 are respectively connected to data 4. Accordingly, the computing device 20 determines to assign a subgroup to each unlabeled data and the data that has a connection relationship with the unlabeled data.

[0235] In another possible implementation, for each unlabeled data item in the data weighted graph, computing device 20 determines j data items having a similarity with the data item greater than or equal to a first threshold as a candidate data set for the data item, wherein the data item in the candidate data set has a connection relationship with each data item in the candidate data set. j is an integer greater than or equal to 1, and the first threshold value can be set based on actual circumstances and is not limited in this embodiment of the present application.

[0236] The computing device 20 allocates a subgroup to each unlabeled data and data having a connection relationship with the unlabeled data.

[0237] Exemplary, reference Figure 7, if the first threshold is 0.8, data 1, data 2, data 3, and data 4 are labeled data, and data 5 and data 6 are unlabeled data. The computing device 20 determines the edge whose similarity is greater than the first threshold value of 0.8 among the five similarities represented by the five edges (edge ​​77, edge 78, edge 79, edge 710, and edge 711) connected to the node 75 representing data 5. For example, the computing device 20 determines that the similarity represented by edge 77 is greater than the first threshold value, and the similarities represented by edges 78, edge 79, edge 710, and edge 711 are all less than the first threshold value of 0.8. Then the computing device 20 determines that the data represented by another node 76 connected to the edge 77 connected to node 75 is data 6. In other words, the computing device 20 determines data 6 through data 5, that is, there is a connection relationship between data 5 and data 6. Similarly, the computing device 20 can determine data 1, data 2, and data 5 through data 6, that is, there is a connection relationship between data 2 and data 5 and data 6 respectively. The computing device 20 can determine data data 2 and data 6 from data 1, that is, data 2 and data 6 are respectively connected to data 1. The computing device 20 can determine data data 1 and data 6 from data 2, that is, data 1 and data 6 are respectively connected to data 2. The computing device 20 can determine data data 4 from data 3, that is, data 3 and data 4 are respectively connected. The computing device 20 can determine data data 3 from data 4, that is, data 4 and data 3 are connected. Accordingly, the computing device 20 determines to allocate a subgroup to each unlabeled data and the data that has a connection relationship with the unlabeled data.

[0238] In the data determined by the above two implementation methods, for each unlabeled data, it is determined that the data having a connection relationship with the unlabeled data includes at least one labeled data. If no labeled data is included, the computing device 20 can select a labeled data from the data weighted graph according to the third preset rule and add it to the data having a connection relationship with the unlabeled data. Alternatively, the computing device 20 can also increase the value of k in the above first possible implementation method or reduce the first threshold in the above second possible implementation method, so that the number of data having a connection relationship can be increased, thereby meeting the requirement that the data having a connection relationship with the unlabeled data includes at least one labeled data.

[0239] By determining the quantum groups that need to be allocated through the above two methods, the demand for the number of quanta during quantum annealing can be reduced, and the number of coupling interactions between each two quantum groups can be reduced, effectively reducing the hardware implementation complexity of quantum annealing.

[0240] The above-mentioned third preset rule can be that the computing device 20 randomly selects a labeled data from the data weighted graph according to certain rules, and adds it to the data that has a connection relationship with the unlabeled data. This embodiment of the present application is not limited to this.

[0241] S6052B: The computing device 20 assigns a quantum ID to the unlabeled data and each determined data to form a quantum group corresponding to the data.

[0242] The method in which the computing device 20 assigns a quantum ID to each determined data for unlabeled data to form a quantum group corresponding to the data can be referred to the description of S6051A above and will not be repeated here.

[0243] For example, according to the first possible implementation of S6051B, if the value of k is 2, data 1, data 2, data 3, and data 4 are labeled data, and data 5 and data 6 are unlabeled data. Figure 10 , Figure 10 It shows that the computing device 20 determines that data 1, data 2, data 5, and data 6 need to be allocated to subgroups. Figure 10 As shown, data 1 corresponds to quantum group 81, data 2 corresponds to quantum group 82, data 5 corresponds to quantum group 85, and data 6 corresponds to quantum group 86.

[0244] S6053B: The computing device 20 determines the longitudinal field applied to each quantum in the quantum group corresponding to the label data.

[0245] The method by which the computing device 20 determines the longitudinal field applied to each quantum in the quantum group corresponding to the labeled data can be referred to the description of S6052A above, which will not be repeated here.

[0246] Exemplary, reference Figure 10 , Figure 10 The quanta in the quantum group corresponding to the label data shown in FIG are represented by hollow circles, and the state of each quantum in the quantum group corresponding to the label data is represented by an arrow. For example, the states of quantum 811 and quantum 812 in quantum group 81 are both upward. Figure 10 The quanta in the quantum group corresponding to the unlabeled data shown in are represented by solid circles, and the states of the quanta in the quantum group corresponding to the unlabeled data are unknown.

[0247] S6054B: The computing device 20 determines a coupling effect between the quantum group corresponding to each piece of unlabeled data having a connection relationship and the quantum group corresponding to the unlabeled data.

[0248] The method for determining the coupling effect between the quantum group corresponding to the data having a connection relationship with each unlabeled data and the quantum group corresponding to the unlabeled data can refer to the description of S6053A above and will not be repeated here.

[0249] Exemplary, reference Figure 10 , Figure 10 The black lines in the figure show the coupling between quantum groups. The quantum groups corresponding to the data connected to data 5 (quantum group 82 and quantum group 86) are coupled to the quantum group 85 corresponding to the unlabeled data 5. The quantum groups corresponding to the data connected to data 6 (quantum group 81, quantum group 82, and quantum group 85) are coupled to the quantum group 86 corresponding to the unlabeled data 6.

[0250] Optionally, S6055B can refer to the description of S6054A above, which will not be repeated here.

[0251] Example 2

[0252] Figure 12 A flow chart showing another method for marking data tags provided by an embodiment of the present application is shown. The method can be applied to Figure 1 The system 10 shown includes the following steps:

[0253] S1201~S1204: Please refer to the above S601-S604.

[0254] S1205: The computing device 20 determines target control parameters based on the data weighted graph. The target control parameters include: a longitudinal field applied to each quantum in a quantum group corresponding to unlabeled data connected to labeled data in the data weighted graph, and a field indicating the coupling effect applied between any two quantum groups.

[0255] When the data weighted graph includes a labeled data (corresponding to the first labeled data in the embodiment of the present application) and an unlabeled data (corresponding to the first unlabeled data in the embodiment of the present application), the above-mentioned target control parameter includes: the longitudinal field applied to the quantum in the quantum group corresponding to the first unlabeled data.

[0256] In scenario 1, each unlabeled data in the data weighted graph is assigned a quantum group, and the similarity between each two data corresponds to the coupling effect between the quantum groups corresponding to the two data.

[0257] Based on scenario 1, the computing device 20 may determine the target control parameters through the following steps:

[0258] S12051A: The computing device 20 assigns a quantum ID to each unlabeled data in the data weighted graph. The quanta identified by the quantum ID constitute a quantum group corresponding to the unlabeled data.

[0259] The process of the computing device 20 assigning a quantum ID to each unlabeled data in the data weighted graph can be referred to the description of S6051A above, and will not be repeated here.

[0260] For example, if s is 6, and among the 6 data, the labeled data are data 1, data 2, data 3, and data 4, and include 4 types of labels, and the unlabeled data are data 5 and data 6. Therefore, the computing device 20 determines to allocate 2 quanta to each unlabeled data. Figure 13 , Figure 13 It shows that the computing device 20 allocates two quanta to each of the unlabeled data 5 and data 6 among the 6 data, and these two quanta constitute the quantum group corresponding to data 5 or data 6. Figure 13 As shown, data 5 corresponds to quantum group 135, and data 6 corresponds to quantum group 136.

[0261] S12052A: The computing device 20 determines the longitudinal field on each quantum in the quantum group corresponding to the unlabeled data connected to the labeled data in the data weighted graph.

[0262] Specifically, for each unlabeled data item in the data weighted graph, computing device 20 determines p labeled data items connected to the unlabeled data item. For labeled data item p0 among the p labeled data items, computing device 20 determines longitudinal field 0 based on the similarity between p0 and the unlabeled data item, the second preset relationship, and the label code of p0.

[0263] Among them, the longitudinal field 0 includes the longitudinal field applied to each quantum in the quantum group corresponding to the above-mentioned unlabeled data. For example, if the quantum group corresponding to the above-mentioned data includes 2 quanta, then the longitudinal field 0 includes a first longitudinal field applied to the first quantum in the quantum group corresponding to the above-mentioned unlabeled data, and a second longitudinal field applied to the second quantum in the quantum group corresponding to the above-mentioned unlabeled data. The first longitudinal field is used to indicate the probability of the spin state of the first quantum, and the second longitudinal field is used to indicate the probability of the spin state of the second quantum. For example, if the label code of p0 is "10", it can correspond to the states of two quanta "up, down", and the similarity between p0 and the unlabeled data is 0.8, then the first longitudinal field determined above is used to indicate that the probability of the state spin of the first quantum in the quantum group corresponding to the unlabeled data is "up" is 0.8, and the second longitudinal field determined above is used to indicate that the probability of the state spin of the first quantum in the quantum group corresponding to the unlabeled data is "down" is 0.8.

[0264] Similarly, the computing device 20 can determine p-1 longitudinal fields based on the remaining p-1 labeled data and the second preset relationship. The computing device 20 superimposes the determined p longitudinal fields to obtain the longitudinal field applied to each quantum in the quantum group corresponding to the unlabeled data.

[0265] The second preset relationship is a preset relationship pre-stored by the computing device 20. The second preset relationship is used to record the corresponding relationship between the similarity and the longitudinal field. The second preset relationship can be represented in a table or any other form, which is not limited in this embodiment of the present application.

[0266] Exemplary, reference Figure 7 , if data 1, data 2, data 3, and data 4 are labeled data, and data 5 and data 6 are unlabeled data. The computing device 20 determines that the labeled data connected to the unlabeled data 5 includes data 1, data 2, data 3, and data 4. The computing device 20 determines the longitudinal field 1 based on the label code of data 1, the similarity between data 1 and data 5, and the second preset relationship. Similarly, the computing device 20 determines the longitudinal field 2 based on data 2 and the second preset relationship, the computing device 20 determines the longitudinal field 3 based on data 3 and the second preset relationship, and the computing device 20 determines the longitudinal field 4 based on data 4 and the second preset relationship. Then, the computing device 20 superimposes the longitudinal field 1, the longitudinal field 2, the longitudinal field 3, and the longitudinal field 4 to obtain the longitudinal field on each quantum in the quantum group corresponding to the unlabeled data 5.

[0267] S12053A. The computing device 20 determines the magnitude of the coupling effect between any two quantum groups.

[0268] The calculation device 20 determines the magnitude of the coupling effect between any two quantum groups, which can be referred to the description of S6053A above, and will not be described in detail here. Figure 13 , Figure 13 The black line in the figure shows the coupling between the two quantum groups.

[0269] In scenario 2, each unlabeled data in the data weighted graph is assigned a corresponding quantum group, where some quantum groups have corresponding coupling effects.

[0270] Based on scenario 2, the computing device 20 may determine the control parameters by following the steps below:

[0271] S12051B, the process of the computing device 20 assigning a quantum ID to each unlabeled data in the data weighted graph can refer to the above S12051A and will not be repeated here.

[0272] S12052B, the process of the computing device 20 determining the longitudinal field on each quantum in the quantum group corresponding to the unlabeled data connected to the labeled data in the data weighted graph can refer to the above S12052A.

[0273] S12053B: The computing device 20 determines, among the quantum groups corresponding to the unlabeled data, the quantum groups that need to be coupled.

[0274] Specifically, for each unlabeled data item, the computing device 20 may refer to the description of S6051B above and determine, through the first possible implementation or the second possible implementation, data that has a connection relationship with the unlabeled data item, i.e., a candidate data set corresponding to the unlabeled data item. The computing device 20 determines that the quantum group corresponding to the unlabeled data item and the quantum group corresponding to each data item in the candidate data set corresponding to the unlabeled data item need to be coupled and connected.

[0275] S12054B, the process of the computing device 20 determining the magnitude of the coupling effect applied between each two quantum groups that need to be coupled and connected as determined in S12053B above, can refer to S12053A above.

[0276] After executing S1205 , the computing device 20 executes S606 , and the quantum annealing device 30 executes S607 . Then, the computing device 20 executes S608 .

[0277] Thus, the data tagging method provided by the embodiments of the present application uses quantum annealing to tag data based on both labeled and unlabeled data. During quantum annealing, only the unlabeled data is assigned a corresponding quantum group for quantum annealing. This not only allows m types of labels to be simultaneously assigned to multiple unlabeled data, improving data tagging efficiency, but also conserves quantum resources and effectively reduces the hardware complexity of the quantum annealing equipment during quantum annealing.

[0278] In summary, the embodiments of the present application provide a method and apparatus for labeling data, which can be applied to a system for labeling data. The system includes a computing device and a quantum annealing device. The computing device determines a target control parameter based on at least one labeled data and at least one unlabeled data, and inputs the target control parameter into the quantum annealing device to obtain the state of the quantum corresponding to the unlabeled data. The label of the data can be characterized by the state of each quantum in the quantum group corresponding to the data. The computing device decodes the quantum state output by the quantum annealing device to obtain the label of the unlabeled data. Compared with the prior art, the method for labeling data provided by the embodiments of the present application can label multiple unlabeled data with different labels at the same time, effectively improving the efficiency of data labeling and reducing the demand for the amount of labeled data.

[0279] The above mainly introduces the solution provided by the embodiment of the present application from the perspective of method. In order to realize the above functions, it includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should easily appreciate that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in a hardware or computer software driven hardware manner depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0280] The embodiment of the present application can divide the computing device and the quantum annealing device into functional modules according to the above method example. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiment of the present application is schematic and is only a logical functional division. There may be other division methods in actual implementation.

[0281] like Figure 14 As shown, Figure 14 The schematic diagram of the structure of the computing device 140 provided in the embodiment of the present application is shown. The computing device 140 is used to execute the above-mentioned method of marking data tags, for example, to execute Figure 6a 、 Figure 6b 、 Figure 6c 、 Figure 6d as well as Figure 12 The computing device 140 may include a determining unit 141 and a sending unit 142 .

[0282] Determining unit 141 is configured to determine target control parameters based on the first labeled data and the first unlabeled data. The target control parameters include: a longitudinal field applied to quanta in the quantum group corresponding to the first unlabeled data; or, alternatively, the target control parameters include: a longitudinal field applied to quanta in the quantum group corresponding to the first labeled data, and a coupling effect between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data. Then, transmitting unit 142 is configured to transmit the target control parameters determined by determining unit 141 to a quantum annealing device. The target control parameters are used to instruct the quantum annealing device to perform quantum annealing to obtain a target annealing result. The target annealing result includes the states of the quanta in the quantum group corresponding to the first unlabeled data. The states of the quanta in the quantum group corresponding to the first unlabeled data are used to represent the label of the first unlabeled data.

[0283] For example, combined with Figure 6a , the determining unit 141 may be used to execute S605, and the sending unit 142 may be used to execute S606. Figure 6c , the determination unit 141 can be used to execute S6051A to S6054A. Figure 6d , the determining unit 141 can be used to execute S6051B to S6054B. Figure 12 The determining unit 141 may be configured to execute S1205 , where S1205 includes S12051A to S12053A, or S1205 includes S12051B to S12054B.

[0284] Optionally, the determining unit 141 is specifically configured to determine the target control parameter based on the similarity between the first labeled data and the first unlabeled data, and an encoding result of the label of the first labeled data.

[0285] For example, combined with Figure 6a , the determining unit 141 can be used to execute S605. Figure 6c , the determination unit 141 can be used to execute S6051A to S6054A. Figure 6d , the determining unit 141 can be used to execute S6051B to S6054B. Figure 12 The determining unit 141 may be configured to execute S1205 , where S1205 includes S12051A to S12053A, or S1205 includes S12051B to S12054B.

[0286] Optionally, if the computing device includes at least two unlabeled data, and the at least two unlabeled data include first unlabeled data, in this case, the target control parameter further includes: a coupling effect applied between quantum groups corresponding to each pair of unlabeled data in the at least two unlabeled data. The target annealing result further includes: the states of quanta in quantum groups corresponding to the unlabeled data other than the first unlabeled data in the at least two unlabeled data.

[0287] Optionally, if the computing device includes at least two labeled data, and the at least two labeled data include the first labeled data, in this case, the determining unit 141 is specifically configured to determine the target control parameter based on a similarity between each labeled data in the at least two labeled data and the first unlabeled data, and encoding results of the labels of the at least two labeled data.

[0288] For example, combined with Figure 6a , the determining unit 141 can be used to execute S605. Figure 6c , the determination unit 141 can be used to execute S6051A to S6054A. Figure 6d , the determining unit 141 can be used to execute S6051B to S6054B. Figure 12 The determining unit 141 may be configured to execute S1205 , where S1205 includes S12051A to S12053A, or S1205 includes S12051B to S12054B.

[0289] Optionally, if the at least two labeled data items have at least two labels, computing device 140 further includes a sorting unit 143 and an encoding unit 144. Sorting unit 143 is configured to sort the labels of the at least two labeled data items based on the similarity between each of the at least two labels to obtain a first sequence. Encoding unit 144 is configured to encode the labels in the first sequence using Gray coding to obtain encoding results of the at least two labels, where the encoding results of the at least two labels include the encoding result of the label of the first labeled data item.

[0290] For example, combined with Figure 6b The sorting unit 143 can be used to execute S6031 to S6033, and the encoding unit 144 can be used to execute S6034.

[0291] Optionally, if the encoding result of a tag is represented by N (N is an integer greater than or equal to 1) bits, the determination unit 141 is specifically configured to determine the nth control parameter based on the similarity between the first tagged data and the first untagged data, and the nth (1≤n≤N, n is an integer) bit in the encoding result of the tag of the first tagged data. Here, the target control parameter includes the first control parameter to the Nth control parameter. Then, the sending unit 142 is specifically configured to send the nth control parameter determined by the determination unit 141 to the quantum annealing device. The nth control parameter is used to instruct the quantum annealing device to perform quantum annealing to obtain an nth annealing result. The nth annealing result includes the state of the quantum in the quantum group corresponding to the first untagged data. Here, the first annealing result to the Nth annealing result are collectively used to represent the tag of the first untagged data.

[0292] For example, combined with Figure 6a , the determining unit 141 may be used to execute S605, and the sending unit 142 may be used to execute S606. Figure 12 , the determining unit 141 can be used to execute S1205.

[0293] Optionally, if the computing device includes Q (Q is an integer greater than or equal to 3) data, the Q data include first labeled data and first unlabeled data. In this case, the computing device 140 also includes an acquisition unit 147. The acquisition unit 147 is used to obtain the similarity between the qth (1≤q≤Q, q is an integer) data in the Q data and the data other than the qth data in the Q data. The determination unit 141 is also used to determine the candidate data set corresponding to the qth data based on the similarity obtained by the acquisition unit 147; the candidate data set corresponding to the qth data includes: data in the Q data whose similarity with the qth data is greater than or equal to a threshold; the qth data has a connection relationship with each data in the candidate data set; the determination unit 141 is also specifically used to determine the target control parameter based on the similarity between the data in the Q data that has a connection relationship with the unlabeled data, and the encoding result of the label of the labeled data in the Q data.

[0294] For example, combined with Figure 6d , the acquiring unit 147 and the determining unit 141 can be used to execute S6051B. Figure 12 , the determining unit 141 can be used to execute S12051B, and the sending unit 142 can be used to execute S606.

[0295] Optionally, the computing device 140 may further include a receiving unit 145. The receiving unit 145 is configured to receive the annealing result sent by the quantum annealing device. Figure 6a , the receiving unit 145 can be used to respond to S607.

[0296] Optionally, the computing device 140 may further include a decoding unit 148. The decoding unit 148 is configured to decode the annealing result received by the receiving unit 145 to obtain a label represented by the annealing result. Figure 6a , the decoding unit 148 can be used to respond to S608.

[0297] Of course, the computing device 140 provided in the embodiment of the present application includes but is not limited to the above units. For example, the computing device 140 may further include a storage unit 146. The storage unit 146 may be used to store program codes of the computing device 140.

[0298] For the detailed description of the above optional manners, please refer to the above method embodiments, which will not be repeated here. In addition, the explanation of any of the computing devices 140 provided above and the description of the beneficial effects can refer to the above corresponding method embodiments, which will not be repeated here.

[0299] As an example, combined with Figure 2The functions implemented by the receiving unit 145, the acquiring unit 147 and the sending unit 142 in the computing device 140 can be realized by Figure 2 The functions implemented by the determining unit 141, the sorting unit 143, the encoding unit 144 and the decoding unit 148 can be realized by Figure 2 Processor 21 in the Figure 2 The functions implemented by the storage unit 146 can be realized by the program code in the memory 22. Figure 2 The memory 22 in is implemented.

[0300] like Figure 15 As shown, Figure 15 The structure diagram of the quantum annealing device 150 provided in the embodiment of the present application is shown. The quantum annealing device 150 is used to perform the above-mentioned method of marking data tags, for example, Figure 6a The quantum annealing device 150 may include a receiving unit 151 , a quantum annealing unit 152 and a sending unit 153 .

[0301] Receiving unit 151 is configured to receive target control parameters sent by a computing device. The target control parameters are determined based on the first tagged data and the first untagged data. The target control parameters include: a longitudinal field applied to quanta in the quantum group corresponding to the first untagged data; or, alternatively, the target control parameters include: a longitudinal field applied to quanta in the quantum group corresponding to the first tagged data, and a coupling effect between the quantum group corresponding to the first tagged data and the quantum group corresponding to the first untagged data. Quantum annealing unit 152 is configured to perform quantum annealing based on the target control parameters received by receiving unit 151 to obtain a target annealing result. The target annealing result includes the states of quanta in the quantum group corresponding to the first untagged data, and the states of quanta in the quantum group corresponding to the first untagged data are used to represent the tag of the first untagged data. Transmitting unit 153 is configured to transmit the target annealing result to the computing device.

[0302] For example, combined with Figure 6a The receiving unit 151 may be used to respond to S606 , the quantum annealing unit 152 may be used to execute S607 , and the sending unit 153 may be used to execute S607 .

[0303] Optionally, if the computing device includes at least two unlabeled data, the at least two unlabeled data include the first unlabeled data. In this case, the target control parameter further includes: a coupling effect applied between quantum groups corresponding to each pair of unlabeled data in the at least two unlabeled data, where the at least two unlabeled data include the first unlabeled data. The target annealing result further includes: the states of quanta in the quantum groups corresponding to the unlabeled data in the at least two unlabeled data, excluding the first unlabeled data.

[0304] Optionally, the quantum annealing unit 152 is specifically configured to perform quantum annealing according to the nth (1≤n≤N, n is an integer, and N is an integer greater than or equal to 1) control parameter to obtain an nth annealing result. The nth control parameter is used to indicate: the longitudinal field of the quantum corresponding to the nth coding bit of the label of the first unlabeled data; or the longitudinal field of the quantum corresponding to the nth coding bit of the label of the first labeled data, and the coupling between the quantum corresponding to the nth coding bit in the coding results of the first labeled data and the label of the first unlabeled data. If a label has N coding bits, the target control parameters include the 1st control parameter to the Nth control parameter. The above-mentioned nth annealing result includes the state of the quantum in the quantum group corresponding to the first unlabeled data, and the 1st annealing result to the Nth annealing result are jointly used to characterize the label of the first unlabeled data.

[0305] For example, combined with Figure 6a , the quantum annealing unit 152 can be used to perform S607.

[0306] Of course, the quantum annealing device 150 provided in the embodiments of the present application includes but is not limited to the above-mentioned units. For example, the quantum annealing device 150 may also include a control circuit unit 154 and a storage unit 155. The control circuit unit 154 is used to apply a longitudinal field to the quanta in the quantum group corresponding to the unlabeled data; alternatively, the control circuit unit 154 is used to apply a longitudinal field to the quanta in the quantum group corresponding to the labeled data, and to apply coupling between the quantum group corresponding to the labeled data and the quantum group corresponding to the unlabeled data. The storage unit 155 can be used to store the program code of the quantum annealing device 150.

[0307] For the detailed description of the above optional methods, please refer to the above method embodiments, which will not be repeated here. In addition, the explanation and description of the beneficial effects of any of the above quantum annealing devices 150 can refer to the above corresponding method embodiments, which will not be repeated here.

[0308] As an example, combined with Figure 3 The functions of the receiving unit 151 and the sending unit 153 in the quantum annealing device 150 can be realized by Figure 3The functions implemented by the quantum annealing unit 152 and the control circuit unit 154 can be realized by Figure 3 The quantum processor 31 in is implemented.

[0309] The embodiment of the present application also provides a chip system 160, such as Figure 16 As shown, the chip system 160 includes at least one processor 161 and at least one interface circuit 162. The processor 161 and the interface circuit 162 can be interconnected via lines. For example, the interface circuit 162 can be used to receive signals from other devices (such as the computing device 140 and the quantum annealing device 150). For another example, the interface circuit 162 can be used to send signals to other devices (such as the processor 161). Exemplarily, the interface circuit 162 can read instructions stored in the memory and send the instructions to the processor 161. When the instructions are executed by the processor 161, the computing device or the quantum annealing device can perform the various steps in the above embodiments. Of course, the chip system 160 can also include other discrete devices, which are not specifically limited in the embodiments of the present application.

[0310] Another embodiment of the present application further provides a computer-readable storage medium, which stores instructions. When the instructions are executed on a computing device or a quantum annealing device, the computing device or the quantum annealing device executes each step performed by the computing device or the quantum annealing device in the method flow shown in the above method embodiment.

[0311] In some embodiments, the disclosed methods may be implemented as computer program instructions encoded in a machine-readable format on a computer-readable storage medium or on other non-transitory media or articles of manufacture.

[0312] Figure 17 A conceptual partial view of a computer program product provided by an embodiment of the present application is schematically shown, wherein the computer program product includes a computer program for executing a computer process on a computing device or a quantum annealing device.

[0313] In one embodiment, the computer program product is provided using a signal bearing medium 170. The signal bearing medium 170 may include one or more program instructions that, when executed by one or more processors, may provide the above-described Figure 6a Thus, for example, reference to Figure 6a One or more features of S601 to S608 may be performed by one or more instructions associated with the signal bearing medium 170. Figure 17 The program instructions in also describe example instructions.

[0314] In some examples, signal-bearing medium 170 may include computer-readable medium 171, such as, but not limited to, a hard drive, a compact disk (CD), a digital video disk (DVD), a digital tape, a memory, a read-only memory (ROM), a random access memory (RAM), and the like.

[0315] In some implementations, signal bearing medium 170 may include computer recordable medium 172 such as, but not limited to, memory, read / write (R / W) CD, R / W DVD, or the like.

[0316] In some embodiments, signal bearing medium 170 may include communication medium 173 such as, but not limited to, digital and / or analog communication media (eg, fiber optic cables, waveguides, wired communication links, wireless communication links, etc.).

[0317] Signal bearing medium 170 may be communicated by a wireless form of communication medium 173 (eg, a wireless communication medium conforming to the IEEE 802.11 standard or other transmission protocols). The one or more program instructions may be, for example, computer executable instructions or logic implemented instructions.

[0318] In some examples, such as for Figure 6a The described computing device or quantum annealing device may be configured to provide various operations, functions, or actions in response to one or more program instructions via computer-readable media 171 , computer-recordable media 172 , and / or communication media 173 .

[0319] Should be understood that the arrangement described here is only for the purpose of example. Thus, those skilled in the art will understand that other arrangements and other elements (such as, machines, interfaces, functions, sequences, and functional groups, etc.) can be used instead, and some elements can be omitted altogether according to the desired result. In addition, many of the described elements can be implemented as discrete or distributed components or in any appropriate combination and position in conjunction with the functional entities implemented by other components.

[0320] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using a software program, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer execution instructions are loaded and executed on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, computer, server or data center to another website, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more media that can be integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium (eg, a solid state disk (SSD)).

[0321] The above is only a specific embodiment of the present application. Those skilled in the art may conceive of changes or substitutions based on the specific embodiment provided in this application, and all such changes or substitutions shall fall within the scope of protection of this application.

Claims

1. A method for labeling data, characterized in that: Applied to a computing device, the method includes: Determining target control parameters based on first labeled data and first unlabeled data; wherein the target control parameters include: a longitudinal field applied to quanta in a quantum group corresponding to the first unlabeled data; or, the target control parameters include: a longitudinal field applied to quanta in a quantum group corresponding to the first labeled data, and a coupling effect applied between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data; The target control parameter is sent to a quantum annealing device, where the target control parameter is used to instruct the quantum annealing device to perform quantum annealing to obtain a target annealing result, where the target annealing result includes a state of a quantum in a quantum group corresponding to the first label-free data, where the state of the quantum in the quantum group corresponding to the first label-free data is used to represent a label of the first label-free data.

2. The method according to claim 1, characterized in that The determining of the target control parameter based on the first labeled data and the first unlabeled data includes: The target control parameter is determined based on the similarity between the first labeled data and the first unlabeled data and an encoding result of the label of the first labeled data.

3. The method according to claim 2, characterized in that The computing device includes at least two labeled data, the at least two labeled data including the first labeled data; and determining the target control parameter based on a similarity between the first labeled data and the first unlabeled data and an encoding result of a label of the first labeled data includes: The target control parameter is determined based on the similarity between each labeled data of the at least two labeled data and the first unlabeled data, and encoding results of the labels of the at least two labeled data.

4. The method according to claim 3, characterized in that The at least two labeled data have at least two labels; the method further includes: sorting the at least two labels with labeled data based on the similarity between each two labels of the at least two labels to obtain a first sequence; Gray coding is used to encode the tags in the first sequence to obtain coding results of the at least two tags; the coding results of the at least two tags include the coding result of the first tag with tagged data.

5. The method according to any one of claims 2 to 4, characterized in that The encoding result of the label of any data in the computing device is represented by N bits, where N is an integer greater than or equal to 1; and the determining the target control parameter based on the similarity between the first labeled data and the first unlabeled data and the encoding result of the label of the first labeled data includes: Determining an nth control parameter based on a similarity between the first labeled data and the first unlabeled data and an nth bit in an encoding result of the label of the first labeled data; wherein the target control parameters include the first control parameter to the Nth control parameter; 1≤n≤N, where n is an integer; The sending the target control parameter to the quantum annealing device includes: The nth control parameter is sent to the quantum annealing device, where the nth control parameter is used to instruct the quantum annealing device to perform quantum annealing to obtain an nth annealing result, where the nth annealing result includes states of quanta in the quantum group corresponding to the first unlabeled data; wherein the first annealing result to the Nth annealing result are collectively used to characterize the label of the first unlabeled data.

6. The method according to claim 1 or 2, characterized in that The computing device includes at least two unlabeled data, the at least two unlabeled data including the first unlabeled data; The target control parameter further includes: a coupling effect applied between quantum groups corresponding to each two unlabeled data in the at least two unlabeled data; The target annealing result further includes: states of quanta in a quantum group corresponding to unlabeled data other than the first unlabeled data among the at least two unlabeled data.

7. The method according to any one of claims 1 to 4, characterized in that The computing device includes Q data, where Q is an integer greater than or equal to 3, and the Q data include the first labeled data and the first unlabeled data; the method further includes: Obtaining similarities between the qth data among the Q data and the data other than the qth data among the Q data, 1≤q≤Q, where q is an integer; Based on the obtained similarity, determining a candidate data set corresponding to the qth data; the candidate data set corresponding to the qth data includes: data among the Q data whose similarity with the qth data is greater than or equal to a threshold; the qth data has a connection relationship with each data in the candidate data set; The determining of the target control parameter based on the first labeled data and the first unlabeled data includes: The target control parameter is determined based on the similarity between the data in the Q data that have a connection relationship with the unlabeled data and the encoding result of the label of the labeled data in the Q data.

8. A method for labeling data, characterized in that: Applied to a quantum annealing device, the method comprises: Receiving a target control parameter sent by a computing device; wherein the target control parameter is determined based on first labeled data and first unlabeled data; the target control parameter includes: a longitudinal field applied to quanta in a quantum group corresponding to the first unlabeled data; or, the target control parameter includes: a longitudinal field applied to quanta in a quantum group corresponding to the first labeled data, and a coupling effect applied between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data; Quantum annealing is performed according to the target control parameter to obtain a target annealing result, and the target annealing result is sent to the computing device; wherein the target annealing result includes the state of the quantum in the quantum group corresponding to the first label-free data, and the state of the quantum in the quantum group corresponding to the first label-free data is used to represent the label of the first label-free data.

9. The method according to claim 8, characterized in that The target control parameter further includes: a coupling effect applied between quantum groups corresponding to each two unlabeled data in at least two unlabeled data; the at least two unlabeled data include the first unlabeled data; The target annealing result further includes: states of quanta in a quantum group corresponding to unlabeled data other than the first unlabeled data among the at least two unlabeled data.

10. The method according to claim 8 or 9, characterized in that The step of performing quantum annealing according to the target control parameters to obtain a target annealing result includes: Perform quantum annealing according to the nth control parameter to obtain the nth annealing result; The nth control parameter is used to indicate: the longitudinal field of the quantum corresponding to the nth coding bit of the label of the first unlabeled data; or the longitudinal field of the quantum corresponding to the nth coding bit of the label of the first tagged data, and the coupling effect between the quantum corresponding to the nth coding bit in the coding results of the first tagged data and the label of the first unlabeled data; The tag of any data has N coding bits; 1≤n≤N, where n is an integer; and N is an integer greater than or equal to 1; The target control parameters include the first control parameter to the Nth control parameter; The nth annealing result includes the state of the quantum in the quantum group corresponding to the first unlabeled data; the first annealing result to the Nth annealing result are commonly used to characterize the label of the first unlabeled data.

11. A computing device, characterized in that The computing device includes a determining unit and a sending unit; The determining unit is configured to determine a target control parameter based on the first labeled data and the first unlabeled data; wherein the target control parameter includes: a longitudinal field applied to quanta in a quantum group corresponding to the first unlabeled data; or the target control parameter includes: a longitudinal field applied to quanta in a quantum group corresponding to the first labeled data, and a coupling effect applied between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data; The sending unit is configured to send the target control parameter determined by the determining unit to a quantum annealing device, where the target control parameter is used to instruct the quantum annealing device to perform quantum annealing to obtain a target annealing result, where the target annealing result includes a state of a quantum in a quantum group corresponding to the first unlabeled data, where the state of the quantum in the quantum group corresponding to the first unlabeled data is used to represent a label of the first unlabeled data.

12. The computing device according to claim 11, wherein: The determining unit is specifically configured to determine the target control parameter based on a similarity between the first labeled data and the first unlabeled data, and an encoding result of a label of the first labeled data.

13. The computing device according to claim 12, wherein: The computing device includes at least two labeled data, the at least two labeled data including the first labeled data; The determining unit is specifically configured to determine the target control parameter based on the similarity between each labeled data of the at least two labeled data and the first unlabeled data, and encoding results of the labels of the at least two labeled data.

14. The computing device according to claim 13, wherein: The at least two labeled data have at least two labels, and the computing device further includes a sorting unit and an encoding unit; The sorting unit is configured to sort the at least two labels with labeled data based on the similarity between each two labels of the at least two labels to obtain a first sequence; The encoding unit is configured to encode the tags in the first sequence using Gray coding to obtain encoding results of the at least two tags; the encoding results of the at least two tags include the encoding result of the first tag with tagged data.

15. The computing device according to any one of claims 12 to 14, characterized in that The encoding result of the label of any data in the computing device is represented by N bits, where N is an integer greater than or equal to 1; The determining unit is specifically configured to determine an nth control parameter based on a similarity between the first labeled data and the first unlabeled data, and an nth bit in an encoding result of the label of the first labeled data; wherein the target control parameters include the first control parameter to the Nth control parameter; 1≤n≤N, where n is an integer; The sending unit is specifically configured to send the nth control parameter determined by the determining unit to the quantum annealing device, where the nth control parameter is used to instruct the quantum annealing device to perform quantum annealing to obtain an nth annealing result, where the nth annealing result includes states of quanta in a quantum group corresponding to the first unlabeled data; wherein the first annealing result to the Nth annealing result are collectively used to represent a label of the first unlabeled data.

16. The computing device according to claim 11 or 12, characterized in that The computing device includes at least two unlabeled data, the at least two unlabeled data including the first unlabeled data; The target control parameter further includes: a coupling effect applied between quantum groups corresponding to each two unlabeled data in the at least two unlabeled data; The target annealing result further includes: states of quanta in a quantum group corresponding to unlabeled data other than the first unlabeled data among the at least two unlabeled data.

17. The computing device according to any one of claims 11 to 14, characterized in that The computing device includes Q data, where Q is an integer greater than or equal to 3, and the Q data include the first labeled data and the first unlabeled data; the computing device also includes an acquisition unit; The acquiring unit is configured to acquire similarities between the qth data among the Q data and the data other than the qth data among the Q data, where 1≤q≤Q, and q is an integer; The determining unit is further configured to determine a candidate data set corresponding to the qth data based on the similarity obtained by the obtaining unit; The candidate data set corresponding to the qth data includes: data among the Q data whose similarity with the qth data is greater than or equal to a threshold; The qth data has a connection relationship with each data in the candidate data set; and is specifically used to determine the target control parameter based on the similarity between the data in the Q data that have a connection relationship with the unlabeled data, and the encoding result of the label of the labeled data in the Q data.

18. A quantum annealing device, characterized in that: The quantum annealing device includes a receiving unit, a quantum annealing unit and a sending unit; The receiving unit is configured to receive a target control parameter sent by a computing device; wherein the target control parameter is determined based on first labeled data and first unlabeled data; the target control parameter includes: a longitudinal field applied to quanta in a quantum group corresponding to the first unlabeled data; or, the target control parameter includes: a longitudinal field applied to quanta in a quantum group corresponding to the first labeled data, and a coupling effect applied between the quantum group corresponding to the first labeled data and the quantum group corresponding to the first unlabeled data; The quantum annealing unit is configured to perform quantum annealing according to the target control parameter received by the receiving unit to obtain a target annealing result; wherein the target annealing result includes the state of the quantum in the quantum group corresponding to the first label-free data, and the state of the quantum in the quantum group corresponding to the first label-free data is used to represent the label of the first label-free data; The sending unit is configured to send the target annealing result to the computing device.

19. The quantum annealing device according to claim 18, characterized in that The target control parameter further includes: a coupling effect applied between quantum groups corresponding to each two unlabeled data in at least two unlabeled data; the at least two unlabeled data include the first unlabeled data; The target annealing result further includes: states of quanta in a quantum group corresponding to unlabeled data other than the first unlabeled data among the at least two unlabeled data.

20. The quantum annealing device according to claim 18 or 19, characterized in that The quantum annealing unit is specifically used to perform quantum annealing according to the nth control parameter to obtain the nth annealing result; The nth control parameter is used to indicate: the longitudinal field of the quantum corresponding to the nth coding bit of the label of the first unlabeled data; or the longitudinal field of the quantum corresponding to the nth coding bit of the label of the first tagged data, and the coupling effect between the quantum corresponding to the nth coding bit in the coding results of the first tagged data and the label of the first unlabeled data; The tag of any data has N coding bits; 1≤n≤N, where n is an integer; and N is an integer greater than or equal to 1; The target control parameters include the first control parameter to the Nth control parameter; The nth annealing result includes the state of the quantum in the quantum group corresponding to the first unlabeled data; the first annealing result to the Nth annealing result are commonly used to characterize the label of the first unlabeled data.

21. A computing device, characterized in that include: memory and one or more processors; The memory is coupled to the processor; The memory is used to store computer program codes, where the computer program codes include computer instructions. When the computer instructions are executed by the computing device, the computing device is enabled to perform the method for marking data tags according to any one of claims 1 to 7.

22. A quantum annealing device, characterized in that: include: memory for one or more quantum processors; The memory is used to store computer program codes, wherein the computer program codes include computer instructions. When the computer instructions are executed by the quantum processor, the quantum annealing device performs the method for marking data tags according to any one of claims 8 to 10.

23. A system for marking data labels, characterized in that: The system includes a computing device and a quantum annealing device, wherein the computing device is used to execute the method for marking data tags according to any one of claims 1 to 7; and the quantum annealing device is used to execute the method for marking data tags according to any one of claims 8 to 10.

24. A chip system, characterized in that: The chip system is applied to a computing device; the chip system includes one or more interface circuits and one or more processors; The interface circuit and the processor are interconnected via a line; the interface circuit is used to receive a signal from the memory of the computing device and send the signal to the processor, wherein the signal includes a computer instruction stored in the memory; when the processor executes the computer instruction, the computing device executes the method for marking data tags as described in any one of claims 1-7.

25. A chip system, characterized in that: The chip system is applied to a quantum annealing device; the chip system includes one or more interface circuits and one or more quantum processors; The interface circuit and the quantum processor are interconnected via a line; the interface circuit is configured to receive a signal from a memory of the quantum annealing device and send the signal to the quantum processor, wherein the signal includes a computer instruction stored in the memory; when the quantum processor executes the computer instruction, the quantum annealing device executes the method for marking a data tag according to any one of claims 8 to 10.

26. A computer-readable storage medium, characterized in that The method comprises computer instructions, which, when executed on a computing device, enable the computing device to implement the method for marking data tags according to any one of claims 1 to 7.

27. A computer-readable storage medium, characterized in that The method comprises computer instructions, which, when executed on a quantum annealing device, enable the quantum annealing device to implement the method for marking data tags according to any one of claims 8 to 10.

Images