CN112561069A - Model processing method, device, equipment, storage medium and product - Google Patents

Model processing method, device, equipment, storage medium and product Download PDF

Info

Publication number
CN112561069A
CN112561069A CN202011548207.4A CN202011548207A CN112561069A CN 112561069 A CN112561069 A CN 112561069A CN 202011548207 A CN202011548207 A CN 202011548207A CN 112561069 A CN112561069 A CN 112561069A
Authority
CN
China
Prior art keywords
quantum circuit
parameterized quantum
parameterized
sample
coding information
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202011548207.4A
Other languages
Chinese (zh)
Other versions
CN112561069B (en
Inventor
王鑫
李广西
陈然一鎏
王友乐
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Baidu Netcom Science and Technology Co Ltd
Original Assignee
Beijing Baidu Netcom Science and Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Baidu Netcom Science and Technology Co Ltd filed Critical Beijing Baidu Netcom Science and Technology Co Ltd
Priority to CN202011548207.4A priority Critical patent/CN112561069B/en
Publication of CN112561069A publication Critical patent/CN112561069A/en
Application granted granted Critical
Publication of CN112561069B publication Critical patent/CN112561069B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N10/00—Quantum computing, i.e. information processing based on quantum-mechanical phenomena
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computational Mathematics (AREA)
  • Condensed Matter Physics & Semiconductors (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Image Analysis (AREA)

Abstract

The disclosure provides a model processing method, a model processing device, model processing equipment, storage media and a product, and relates to the field of quantum computing. The specific implementation scheme is as follows: acquiring coding information of a parameterized quantum circuit sample to be trained; calculating a second moment of a parameter gradient of a parameterized quantum circuit sample to be trained; taking the second moment of the parameter gradient of the parameterized quantum circuit sample as a label, and labeling the coding information of the parameterized quantum circuit sample to obtain labeled coding information; and inputting the marked coding information into a preset neural network for model training, and obtaining a target neural network after the training is finished, wherein the target neural network can predict and obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted based on the input coding information of the parameterized quantum circuit to be predicted. In this way, the characteristics of the parameterized quantum circuit can be effectively evaluated.

Description

Model processing method, device, equipment, storage medium and product
Technical Field
The present disclosure relates to the field of data processing technology, and more particularly, to the field of quantum computing.
Background
One very important component in Quantum computing is a Variational Quantum feature solver (VQE), which can extract the ground state information of a physical system using a Noisy Intermediate-Scale Quantum (NISQ) computer. In particular, the effect of the ground state information obtained by the solution using the variational quantum feature solver is closely related to the parameterized quantum circuit given in the solution process, which may affect the accuracy of the finally prepared ground state information and the required time (i.e., the required quantum resources). Therefore, how to evaluate the parameterized quantum circuit to select a suitable parameterized quantum circuit to reduce the required quantum resources becomes an urgent problem to be solved.
Disclosure of Invention
The present disclosure provides a model processing method, apparatus, device, storage medium and product for parameterized quantum circuits.
According to an aspect of the present disclosure, there is provided a model processing method applied to a parameterized quantum circuit, including:
acquiring coding information of a parameterized quantum circuit sample to be trained;
calculating a second moment of a parameter gradient of a parameterized quantum circuit sample to be trained;
taking the second moment of the parameter gradient of the parameterized quantum circuit sample as a label, and labeling the coding information of the parameterized quantum circuit sample to obtain labeled coding information;
and inputting the marked coding information into a preset neural network for model training, and obtaining a target neural network after the training is finished, wherein the target neural network can predict and obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted based on the input coding information of the parameterized quantum circuit to be predicted.
According to another aspect of the present disclosure, there is provided a model processing apparatus applied to a parameterized quantum circuit, including:
the encoding information acquisition unit is used for acquiring encoding information of a parameterized quantum circuit sample to be trained;
the calculation unit is used for calculating the second moment of the parameter gradient of the parameterized quantum circuit sample to be trained;
the sample data processing unit is used for taking the second moment of the parameter gradient of the parameterized quantum circuit sample as a label and marking the coding information of the parameterized quantum circuit sample to obtain marked coding information;
and the model training unit is used for inputting the marked coding information into a preset neural network for model training, and obtaining a target neural network after the training is finished, wherein the target neural network can predict and obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted based on the input coding information of the parameterized quantum circuit to be predicted.
According to another aspect of the present disclosure, there is provided an electronic device including:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a method according to any one of the embodiments of the present disclosure.
According to another aspect of the present disclosure, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform a method in any of the embodiments of the present disclosure.
According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the method in any of the embodiments of the present disclosure.
According to the technology disclosed by the invention, the characteristics of the parameterized quantum circuit can be effectively evaluated, and a foundation is laid for selecting the parameterized quantum circuit meeting the requirements.
It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present disclosure, nor do they limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description.
Drawings
The drawings are included to provide a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:
FIG. 1 is a flow diagram of a model processing method applied to parameterized quantum circuits in accordance with an embodiment of the disclosure;
FIG. 2 is a first flow diagram of a model processing method applied to parameterized quantum circuits in accordance with an embodiment of the present disclosure in a specific example;
FIG. 3 is a flow diagram illustration two in a specific example of a model processing method applied to parameterized quantum circuits in accordance with an embodiment of the present disclosure;
FIG. 4 is a schematic diagram of a parameterized quantum circuit in a specific example of a model processing method applied to the parameterized quantum circuit in accordance with an embodiment of the present disclosure;
FIG. 5 is a schematic block diagram of a model processing device applied to parameterized quantum circuits in accordance with an embodiment of the present disclosure;
fig. 6 is a block diagram of an electronic device for implementing the model processing method applied to parameterized quantum circuits of an embodiment of the present disclosure.
Detailed Description
Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of the embodiments of the disclosure are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
In the disciplines of physics and chemistry, a very important problem is to extract information of Ground State (Ground State) of physical systems such as molecules and atoms, which is called Ground State information for short. Generally, the ground state of a physical system is determined by the Hamiltonian (Hamiltonian) of the physical system. The Hamiltonian H is mathematically in the form of a Hermitian matrix (Hermitian), for example, if the physical system is composed of n qubits (Qubits), the Hamiltonian H is 2n×2nA hermitian matrix, i.e. a complex matrix of conjugate symmetry. Specifically, the ground state of the physical system is the eigenvector of which the hamilton H has the smallest eigenvalue, and based on this, information for extracting the ground state of the physical system is converted into the eigenvector of which the hamilton H has the smallest eigenvalue for solving.
In view of the above problems, a very important component in Quantum computing is a Variational Quantum feature solver (VQE), which can extract a ground state of a physical system by using a Noisy Intermediate-Scale Quantum (NISQ) computer, and it can be understood that the Variational Quantum feature solver is an algorithm in the Noisy Intermediate-Scale Quantum computer, that is, the Variational Quantum feature solver in the Noisy Intermediate-Scale Quantum computer can extract the ground state of the physical system. Specifically, a given Parameterized Quantum Circuit (PQC) is trained using a noisy mesoscale Quantum computer such that the trained Parameterized Quantum Circuit is used to efficiently prepare the ground state of the hamiltonian. However, recent studies have shown that variational quantum feature solvers present a bottleneck: as the scale of physical systems increases (e.g., the number of qubits increases), the resources consumed by computations increase exponentially. This bottleneck greatly limits the computational efficiency and application of the variable-fraction quantum-signature solver. In particular, the effect of the variational quantum feature solver is closely related to the given parameterized quantum circuit, which may affect the accuracy of the final prepared ground state and the time required for the training process (i.e., the required quantum resources), based on which the above-mentioned bottlenecks are also closely related to the parameterized quantum circuit used in the computation. Therefore, the selection of suitable parameterized quantum circuits to reduce the quantum resources required for computation is becoming an important step in the recent development of quantum computation.
However, in the existing scheme, it is very difficult to find the optimal parameterized quantum circuit, because the search space becomes exponentially larger with the increase of the physical system scale, which greatly increases the search difficulty and limits the efficiency. On the other hand, for the variational quantum feature solver, the existing scheme is very dependent on a specific physical system, that is, the optimal parameterized quantum circuit found by the existing scheme can only be used for extracting the features of a specific physical system, and this also limits the universality of the existing scheme.
Based on the above, the scheme of the application provides a model processing method, device, equipment, storage medium and product applied to a parameterized quantum circuit, which can not only evaluate the characteristics of the parameterized quantum circuit, but also judge whether a ground state aiming at a physical system and meeting the precision requirement is effectively prepared, and can save quantum resources required by calculation. Therefore, the scheme of the application plays an important role in exploring the application of the noisy medium-sized quantum computer in the fields of physics, chemistry, machine learning and the like.
Specifically, fig. 1 is a schematic flow diagram of a model processing method applied to a parameterized quantum circuit according to an embodiment of the present disclosure, as shown in fig. 1, the method includes:
step S101: and acquiring the coding information of the parameterized quantum circuit sample to be trained. Here, the parameterized quantum circuit samples to be trained may be randomly sampled from all the parameterized quantum circuits.
Step S102: and calculating the second moment of the parameter gradient of the parameterized quantum circuit sample to be trained.
Step S103: and taking the second moment of the parameter gradient of the parameterized quantum circuit sample as a label, and labeling the coding information of the parameterized quantum circuit sample to obtain the labeled coding information.
Step S104: and inputting the marked coding information into a preset neural network for model training, and obtaining a target neural network after the training is finished, wherein the target neural network can predict and obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted based on the input coding information of the parameterized quantum circuit to be predicted.
Therefore, the characteristics of the parameterized quantum circuit can be effectively evaluated, and a foundation is laid for selecting the parameterized quantum circuit meeting the requirements. Further, the scheme of the application can successfully avoid the problems of the existing scheme, for example, the scheme of the application does not need to search the parameterized quantum circuit in an exponential space and does not depend on a specific physical system, and because the target neural network obtained by training of the scheme of the application can effectively evaluate the parameterized quantum circuit, a foundation is laid for subsequently and efficiently judging whether the parameterized quantum circuit can save computing resources, and the method has strong universality, high efficiency and adaptability.
In a specific example of the scheme of the present application, before performing model training, the following manner may be adopted to obtain encoding information of the parameterized quantum circuit sample, specifically, determine circuit characteristics of the parameterized quantum circuit sample, such as the type of the quantum gate, the position of the quantum gate, and the like; and further, based on the circuit characteristics of the parameterized quantum circuit samples, encoding the parameterized quantum circuit samples to obtain the encoding information of the parameterized quantum circuit samples. Therefore, the encoding information can be used for uniquely determining the parameterized quantum circuit sample, so that the foundation is laid for fundamentally avoiding searching the parameterized quantum circuit in an exponential space, and meanwhile, the foundation is laid for efficiently judging whether the parameterized quantum circuit can save computing resources in the follow-up process.
In a specific example of the present application, the following method may be adopted to determine the circuit characteristics of the parameterized quantum circuit sample, specifically including: determining the position of a target quantum gate in the parameterized quantum circuit sample; and at least using the position of the target quantum gate in the parameterized quantum circuit sample as the circuit characteristic of the parameterized quantum circuit sample. That is to say, the parameterized quantum circuit can be encoded based on the position of the target quantum gate in the parameterized quantum circuit, so that the encoded information can uniquely represent one parameterized quantum circuit, and a foundation is laid for obtaining a mapping relation between the encoded information and the features (such as the second moment of a parameter gradient) of the parameterized quantum circuit by utilizing neural network learning subsequently.
Here, in the actual encoding process, the target quantum gate at a specific position may be set and encoded according to a preset sequence, so as to obtain encoding information for the parameterized quantum circuit.
In a specific example of the present disclosure, the parameterized quantum circuit sample further includes other quantum gates besides the target quantum gate, and positions of the other quantum gates in the parameterized quantum circuit sample are not changed. That is to say, in the training sample, when the training sample is a type of parameterized quantum circuit and the position of one or some quantum gates in the type of parameterized quantum circuit is the same, at this time, encoding can be performed only for the quantum gates having the difference, thereby simplifying the encoding content and improving the encoding efficiency.
For example, as shown in fig. 4, the specific encoding process includes: assuming that the parameterized quantum circuit contains only single-bit revolving gates RyUnit array I and CNOT gate (i.e. Control-X gate), and have fixed circuit template, for example, the position of CNOT gate in the parameterized quantum circuit is fixed, and the quantum gates at other specific positions (as indicated by the square box in fig. 4) can be changed, and in this case, the fixed-position quantum gates, i.e. the CNOT gates, can not be considered in the encoding process. Further, a single bit revolving gate RyThe unit arrays I are arranged in a preset sequence, such as from left to right, from top to bottom, and the like, wherein R isyCoded as 1 and I coded as 0, thereby forming a binary string, i.e., the coded information of the present example.
In a specific example of the present disclosure, after the encoding information is determined, the number of training samples may be further extended based on the encoding information, so as to improve training efficiency, specifically, the parameterized quantum circuit samples are determinedAfter the coding information is coded, repeatedly setting the coding information according to a preset mode, and repeating the preset times to obtain new coding information; and obtaining a new parameterized quantum circuit based on the new coding information so as to take the new parameterized quantum circuit as a parameterized quantum circuit sample to be trained. For example, the encoded information is [1,0,1, 0]]Wherein, the digits in the coded information all correspond to the positions of the parametric quantum circuit, and 1 represents that a single bit R exists at the corresponding positiony(theta) a revolving gate, 0 representing no single bit R at the corresponding positiony(theta) a revolving door. At this time, the process is repeated once to obtain new coded information [1,0,1,0,0, 0]Based on the method, a new parameterized quantum circuit can be obtained, namely, the original parameterized circuit is repeated once, so that the new parameterized quantum circuit can be obtained. Therefore, the number of samples is rapidly increased, and the efficiency of model training is improved.
In a specific example of the scheme of the present application, during the model training, the following steps may be further performed: specifically, determining a loss function, wherein the loss function is determined based on a Hamiltonian of a preset physical system and a parameterized quantum circuit sample acting on the preset physical system; for example, the loss function:
Figure BDA0002856284500000071
where U (θ) is the parameterized quantum circuit and ρ is the initial quantum state of the input parameterized quantum circuit, typically |0><0, H is the hamiltonian of a given physical system (e.g., a quantum system). Further, after the labeled coding information is input to a preset neural network for model training, the loss function can be minimized by adjusting parameters in the parameterized quantum circuit samples, so as to obtain the target neural network. Therefore, a foundation is laid for subsequently and efficiently judging whether the parameterized quantum circuit can save the computing resources.
In a specific example of the scheme of the application, after a target neural network is obtained through training, the target neural network can be used for evaluating the characteristics of any parameterized quantum circuit, and specifically, the parameterized quantum circuit to be predicted is obtained; inputting the parameterized quantum circuit to be predicted to the target neural network to obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted; selecting the parameterized quantum circuit with the second moment estimated value of the parameter gradient larger than a preset value from the parameterized quantum circuit to be predicted as an initial parameterized quantum circuit; determining a ground state of the predetermined physical system using at least the initial parameterized quantum circuit. In this way, the basic state information of the physical system can be obtained quickly, and the initial parametric quantum circuit used for calculating the basic state information is obtained after screening, so that the requirements of high accuracy and less quantum resource use can be maximized.
In a specific example of the scheme of the present application, in order to obtain a parameterized quantum circuit meeting the precision requirement through more precise screening, the following steps may be further performed: selecting a target ground state meeting the preset precision requirement from the obtained ground states of the preset physical system; and further taking the initial parametric quantum circuit corresponding to the target ground state as a target parametric quantum circuit. Therefore, on the basis that the ground state information meeting the precision requirement can be obtained, the computing resources are saved to the maximum extent.
In conclusion, the scheme of the application is different from the prior scheme that the problem of the search circuit and the problem to be solved are coupled together, but the process of determining the parameterized quantum circuit is independent, firstly, the trained target neural network is used to evaluate the characteristics of the parameterized quantum circuit, then the parameterized quantum circuit is searched and the problem to be solved is verified based on the evaluation result, for example, firstly, the proper target parameterized quantum circuit is selected based on the evaluation result, then the ground state information is solved based on the target parameterized quantum circuit, the circuit searching problem and the problem to be solved are processed separately, therefore, the scheme of the application can be more flexibly applied to eigenvalue solution of a wider variety of physical systems (such as quantum chemical systems) and other problems in the field of quantum machine learning, and has more practicability and flexibility.
Compared with the prior art that only the parameterized quantum circuit which effectively solves the problem is simply considered and searched, the method and the device not only consider effectiveness, but also consider the characteristic that when the second moment of the parameter gradient of the parameterized quantum circuit is larger, the used resource is smaller, so that less resources can be used in the quantum system feature solution, and the resources are saved.
The following describes the present application in further detail with reference to specific examples, and specifically, the present application is designed to save resources occupied by a variable component quantum feature solver (VQE). In particular, the present solution makes full use of neural networks widely used in machine learning, and the function of the neural network is to determine whether a given parameterized quantum circuit can save resources. In summary, the solution of the present application is to train a neural network on some given parameterized quantum circuit with reference to the second moment of the parameter gradient of the parameterized quantum circuit, and find out a parameterized quantum circuit that can save resources and is suitable for VQE based on the trained neural network. Specifically, the scheme of the application firstly carries out encoding processing on the prepared parameterized quantum circuit; then randomly sampling a plurality of coded information, wherein each coded information represents a parameterized quantum circuit; calculating the second moment of the parameter gradient of the sampled parametric quantum circuit; respectively taking the coding information obtained by sampling and the second moment of the corresponding parameter gradient as a training sample and a label, and inputting the training sample and the label into a neural network for training; and finally, searching out a parameterized quantum circuit which can save computing resources and can be used in a variational quantum feature solver by using the trained neural network, wherein fewer quantum resources are used in the solving process by using the variational quantum feature solver.
The method is characterized in that firstly, the neural network is adopted to assist in searching to obtain the parameterized quantum circuit, and the parameterized quantum circuit is not selected simply by artificial experience; secondly, the scheme of the application is based on the circuit characteristics of the parameterized quantum circuit to encode the parameterized quantum circuit innovatively, and then the second moment of the parameter gradient of the parameterized quantum circuit can be directly predicted only by knowing the encoding information of the parameterized quantum circuit.
Further, the working principle of the scheme of the application is as follows:
the quantum resources occupied by the variational quantum feature solver are closely related to the second moment of the parameter gradient of the parameterized quantum circuit, that is, the smaller the second moment of the parameter gradient, the more quantum resources are needed, and conversely, the fewer the quantum resources are needed. Based on this, the scheme of the application can ensure that the quantum resources occupied in the solving process are less as long as the second moment of the parameter gradient of the selected parameterized quantum circuit is ensured to be larger. In addition, if each parameterized quantum circuit is properly encoded based on the circuit structure, a mapping relationship between the encoding information and the second moment of the parameter gradient of the parameterized quantum circuit can be obtained, and therefore, by using the characteristic, the neural network can be trained to learn the mapping relationship.
As shown in fig. 2 and 3, the specific process includes:
step 1: given the Hamiltonian H of a physical system (such as a quantum system), the loss function is obtained:
Figure BDA0002856284500000091
where U (θ) is the parameterized quantum circuit and ρ is the initial quantum state of the input parameterized quantum circuit, typically |0><0|。
It should be noted that the hamiltonian H is a hamiltonian of a general physical system, in other words, the present application does not limit a given hamiltonian, that is, the physical system is not limited.
Step 2: the parameterized quantum circuit is encoded and the encoded information (i.e., a binary code, or binary codeword) is denoted as code (c). For example, taking binary coding as an example, if the parameterized quantum circuit consists of a number of single bits Ry(theta) turnstile and CNOT door, according toEach position in the parameterized quantum circuit has no Ry(θ) rotating the gate to determine the encoded information for the parameterized quantum circuit, for example, the encoded information may be similarly [1,0,1,0]Wherein 1 represents yes and 0 represents no, each digit in the coded information is uniquely corresponding to a position, namely the digits are in one-to-one correspondence with the positions of the parameterized quantum circuit, so that the parameterized quantum circuit can be uniquely determined by using given coded information.
And step 3: according to the information coded above, S parametric quantum circuits are randomly sampled, and the adopted parametric quantum circuits are recorded as a set
Figure BDA0002856284500000092
Step 4, calculating the second moment of the parameter gradient of the S parametric quantum circuits by adopting a direct calculation method, and recording the calculation result as
Figure BDA0002856284500000093
Here, the direct calculation method means that first of all, the [0,2 π]The above uniform distribution samples values of parameters of a plurality of (for example, 100) parameterized quantum circuits, and then the gradient of the parameters is calculated according to the loss function L (θ) described above, so as to calculate the second moment of the gradient, that is, the second moment directly calculated by the scheme of the present application.
And 5: and respectively taking the coding information of the S parametric quantum circuits and the second moment of the parameter gradient thereof as an input and a label, and inputting the input and the label to a preset neural network for training. Here, the second moment of the parameter gradient of the parameterized quantum circuit may be used as a label to label the encoded information of the parameterized quantum circuit, and the labeled encoded information (i.e., the encoded information of the second moment carrying the parameter gradient) is used as an input and is input into the preset neural network. Adjusting the state information of the parameterized quantum circuit by adjusting the parameters of the S parameterized quantum circuits to minimize the loss function in step 1; and then finishing training after the loss function in the step 1 is minimized or converged to obtain the target neural network.
Here, the preset neural network may be selected simply, for example, a neural network including 2 hidden layers and an activation function Sigmoid, which is not limited in the present embodiment.
Step 6: binary codes (namely coding information) of all parameterized quantum circuits to be evaluated can be recorded as codes (c) and are input into a target neural network obtained after training is completed, and an output result is obtained, namely, a second-order moment estimated value of the parameter gradient of each parameterized quantum circuit to be evaluated, which is obtained through estimation, and can be recorded as { mon (c) }.
And 7: and (c) selecting the target second-order moment estimated value which meets a certain condition and has a larger second-order moment estimated value as the selected target second-order moment estimated value from the { min (c) } obtained in the step 6, obtaining a binary code corresponding to the target second-order moment estimated value, and recording the binary code as { code (max) }.
And 8: and restoring the corresponding parameterized quantum circuit according to { code (max) } and carrying out ground state solution on the Hamilton quantity H corresponding to the loss function based on the algorithm of the variable-component quantum feature solver so as to verify the parameterized quantum circuit. And further, obtaining a parameterized quantum circuit of which the solving result meets the preset precision requirement, namely the target parameterized quantum circuit expected to be output.
Here, in practical applications, the parameterized quantum circuit may be binary-coded in the following manner, as shown in fig. 4, and the specific coding flow includes:
step 1: assuming that the parameterized quantum circuit contains only single-bit revolving gates RyThe unit array I and the CNOT gate (i.e. Control-X gate), and has a fixed circuit template, for example, the CNOT gate in the parameterized quantum circuit has a fixed position, and the quantum gates in other specific positions can be changed, in which case, the quantum gates in fixed positions, i.e. the CNOT gate, can not be considered in the encoding process.
Step 2: single bit revolving door RyThe unit arrays I are arranged in a preset sequence, such as from left to right, from top to bottom, and the like, wherein R isyCoded as 1 and I coded as 0, thereby forming a binary string, i.e., the coded information of the present example.
Here, it should be noted that, in the quantum circuit shown in fig. 4, the corresponding quantum gate in the frame may be changed, for example, from RyBecome, or from becoming RyAfter transformation, a new parameterized quantum circuit can be obtained. Of course, this example is only used to explain the encoding process of the present application, and is not used to limit the structure and encoding process of the parameterized quantum circuit used in the present application, and in practical applications, other structures or other encoding modes may also be used, and the present application does not limit this.
Further, in order to increase the number of samples, the circuit template shown in fig. 4 may be repeated a predetermined number of times, and in this case, the obtained binary code may be followed by repeated information to obtain new code information. Here, assuming that the repetition is repeated at most 10 times, the repetition number may be expressed by an encoding method, for example, when the repetition number is expressed by one-hot vector of 10 dimensions, for example, the D-th position is set to 1, and the other positions are set to 0, repetition information is obtained, for example, when D is 2, the repetition information of the repetition number is [ 0100000000 ], and thus, the encoding information of the parameterized quantum circuit repeated by the preset number is obtained by adding the encoding information of the parameterized quantum circuit corresponding to the number.
Based on this, for a parameterized quantum circuit with 6 qubits, the final encoded information is a 20-dimensional vector. For example, the coding information of the new parameterized quantum circuit obtained after repeating the parameterized quantum circuit in fig. 4 for 3 times is [ 01110100100010000000 ], where the first 10 items represent coding information and the last ten items represent repetition information.
Therefore, the encoding process of the parameterized quantum circuit is completed, and here, it should be noted that in practical application, other encoding modes can also be adopted, and the scheme of the application is not limited to this, as long as the encoding is performed based on the circuit characteristics of the parameterized quantum circuit, and the parameterized quantum circuit can be uniquely determined by using the encoding.
Therefore, the scheme of the application makes full use of the thought of neural network and machine learning, designs the estimation scheme of the second moment of the parameter gradient of the parameterized quantum circuit, and efficiently and practically detects the trainable performance of the parameterized quantum circuit and the practicability of the parameterized quantum circuit in the variational quantum feature solver, so that the advantages and disadvantages of the parameterized quantum circuit designed on recent quantum equipment can be effectively measured, and further the quantum feature solution is more efficient.
In summary, the scheme of the present application is advantageous in the following aspects:
first, compared to the conventional method, i.e. the method of directly calculating the gradient to obtain the second moment thereof, the scheme of the present application is more efficient because the consumption of the multi-bit quantum state by gradient sampling in the existing scheme increases at a high speed. The scheme of the application can directly judge the characteristics of the parameterized quantum circuit according to the coding information of the quantum circuit, and further provides efficient and practical parameterized quantum circuit selection for quantum characteristic solvers and other quantum machine learning applications.
Second, compared with the conventional method, the scheme of the application has higher practicability. At present, the cost of sampling and gradient calculation on quantum equipment is very expensive, so that the scheme of the application avoids sampling on the quantum equipment, all estimation is completed on a classical computer, and the practicability of the scheme is effectively ensured.
Thirdly, the scheme of the application has better effectiveness. The parameterized quantum circuit found by the scheme can be ensured to be suitable for the variational quantum feature solver, and occupies as few quantum resources as possible.
Fourthly, the scheme of the application has expansibility, and can be widely applied to other quantum computing applications needing to use the parameterized quantum circuit, such as application scenarios of variable component sub-classifiers, quantum approximation optimization and the like.
The present application further provides a model processing apparatus applied to a parameterized quantum circuit, as shown in fig. 5, including:
the encoding information acquiring unit 501 is configured to acquire encoding information of a parameterized quantum circuit sample to be trained;
a calculating unit 502, configured to calculate a second moment of a parameter gradient of a parameterized quantum circuit sample to be trained;
the sample data processing unit 503 is configured to use a second moment of the parameter gradient of the parameterized quantum circuit sample as a tag, and label the coding information of the parameterized quantum circuit sample to obtain labeled coding information;
the model training unit 504 is configured to input the labeled coding information to a preset neural network for model training, and obtain a target neural network after the training is completed, where the target neural network is capable of predicting a second moment estimation value of a parameter gradient of the parameterized quantum circuit to be predicted based on the input coding information of the parameterized quantum circuit to be predicted.
In a specific example of the scheme of the present application, the encoding information obtaining unit is further configured to determine a circuit characteristic of a parameterized quantum circuit sample; and based on the circuit characteristics of the parameterized quantum circuit samples, carrying out coding processing on the parameterized quantum circuit samples to obtain the coding information of the parameterized quantum circuit samples.
In a specific example of the scheme of the present application, the encoding information obtaining unit is further configured to determine a position of a target quantum gate in the parameterized quantum circuit sample; and at least using the position of the target quantum gate in the parameterized quantum circuit sample as the circuit characteristic of the parameterized quantum circuit sample.
In a specific example of the present disclosure, the parameterized quantum circuit sample further includes other quantum gates besides the target quantum gate, and positions of the other quantum gates in the parameterized quantum circuit sample are not changed.
In a specific example of the scheme of the present application, the method further includes:
the training data acquisition unit is used for repeatedly setting the coding information according to a preset mode after the coding information of the parameterized quantum circuit sample is determined, and repeating the preset times to obtain new coding information; and obtaining a new parameterized quantum circuit based on the new coding information so as to take the new parameterized quantum circuit as a parameterized quantum circuit sample to be trained.
In a specific example of the scheme of the present application, the model training unit is further configured to determine a loss function, where the loss function is determined based on a hamiltonian of a preset physical system and a parameterized quantum circuit sample acting on the preset physical system; minimizing the loss function by adjusting parameters in the parameterized quantum circuit samples to obtain the target neural network.
In a specific example of the scheme of the present application, the method further includes:
the model prediction unit is used for acquiring a parameterized quantum circuit to be predicted; inputting the parameterized quantum circuit to be predicted to the target neural network to obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted;
the base state processing unit is used for selecting the parameterized quantum circuit with the second moment estimated value of the parameter gradient larger than a preset value from the parameterized quantum circuit to be predicted as an initial parameterized quantum circuit; determining a ground state of the predetermined physical system using at least the initial parameterized quantum circuit.
In a specific example of the scheme of the application, the ground state processing unit is further configured to select a target ground state meeting a preset precision requirement from the obtained ground states of the preset physical system; and taking the initial parameterized quantum circuit corresponding to the target ground state as a target parameterized quantum circuit.
The functions of each unit in the model processing apparatus according to the embodiment of the present invention may refer to the corresponding descriptions in the above method, and are not described herein again.
Here, it should be noted that the model processing apparatus according to the present disclosure may be a classical device, such as a classical computer, a classical electronic device, and the like, in which case, the above units may be implemented by hardware of the classical device, such as a memory, a processor, and the like. Of course, the model processing apparatus according to the present invention may also be a quantum device, and in this case, each unit may be implemented by quantum hardware or the like.
The present disclosure also provides an electronic device, a readable storage medium, and a computer program product according to embodiments of the present disclosure.
FIG. 6 illustrates a schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 6, the apparatus 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM)602 or a computer program loaded from a storage unit 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 can also be stored. The calculation unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.
A number of components in the device 600 are connected to the I/O interface 605, including: an input unit 606 such as a keyboard, a mouse, or the like; an output unit 607 such as various types of displays, speakers, and the like; a storage unit 608, such as a magnetic disk, optical disk, or the like; and a communication unit 609 such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information/data with other devices via a computer network such as the internet and/or various telecommunication networks.
The computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various dedicated Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, and so forth. The calculation unit 601 performs the various methods and processes described above, such as a model processing method applied to a parameterized quantum circuit. For example, in some embodiments, the model processing method applied to the parameterized quantum circuit may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 600 via the ROM 602 and/or the communication unit 609. When the computer program is loaded into RAM 603 and executed by the computational unit 601, one or more steps of the model processing method described above as applied to a parameterized quantum circuit may be performed. Alternatively, in other embodiments, the computational unit 601 may be configured by any other suitable means (e.g., by means of firmware) to perform the model processing method applied to the parameterized quantum circuit.
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), system on a chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.
The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present disclosure may be executed in parallel or sequentially or in different orders, and are not limited herein as long as the desired results of the technical solutions disclosed in the present disclosure can be achieved.
The above detailed description should not be construed as limiting the scope of the disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present disclosure should be included in the scope of protection of the present disclosure.

Claims (19)

1. A model processing method applied to a parameterized quantum circuit, comprising:
acquiring coding information of a parameterized quantum circuit sample to be trained;
calculating a second moment of a parameter gradient of a parameterized quantum circuit sample to be trained;
taking the second moment of the parameter gradient of the parameterized quantum circuit sample as a label, and labeling the coding information of the parameterized quantum circuit sample to obtain labeled coding information;
and inputting the marked coding information into a preset neural network for model training, and obtaining a target neural network after the training is finished, wherein the target neural network can predict and obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted based on the input coding information of the parameterized quantum circuit to be predicted.
2. The method of claim 1, further comprising:
determining circuit characteristics of the parameterized quantum circuit sample;
and based on the circuit characteristics of the parameterized quantum circuit samples, carrying out coding processing on the parameterized quantum circuit samples to obtain the coding information of the parameterized quantum circuit samples.
3. The method of claim 2, wherein the determining circuit characteristics of the parameterized quantum circuit samples comprises:
determining the position of a target quantum gate in the parameterized quantum circuit sample;
and at least using the position of the target quantum gate in the parameterized quantum circuit sample as the circuit characteristic of the parameterized quantum circuit sample.
4. The method of claim 3, wherein the parameterized quantum circuit sample further comprises other quantum gates in addition to the target quantum gate, the other quantum gates being invariant in position within the parameterized quantum circuit sample.
5. The method of any of claims 2 to 4, further comprising:
after the coding information of the parameterized quantum circuit sample is determined, the coding information is repeatedly set according to a preset mode and is repeated for a preset number of times to obtain new coding information;
and obtaining a new parameterized quantum circuit based on the new coding information so as to take the new parameterized quantum circuit as a parameterized quantum circuit sample to be trained.
6. The method of claim 1, further comprising:
determining a loss function, wherein the loss function is determined based on a Hamiltonian of a preset physical system and a parameterized quantum circuit sample acting on the preset physical system;
after the labeled coding information is input to a preset neural network for model training, the method further comprises the following steps:
minimizing the loss function by adjusting parameters in the parameterized quantum circuit samples to obtain the target neural network.
7. The method of claim 1 or 6, further comprising:
obtaining a parameterized quantum circuit to be predicted;
inputting the parameterized quantum circuit to be predicted to the target neural network to obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted;
selecting the parameterized quantum circuit with the second moment estimated value of the parameter gradient larger than a preset value from the parameterized quantum circuit to be predicted as an initial parameterized quantum circuit;
determining a ground state of a predetermined physical system using at least the initial parameterized quantum circuit.
8. The method of claim 7, further comprising:
selecting a target ground state meeting the preset precision requirement from the obtained ground states of the preset physical system;
and taking the initial parameterized quantum circuit corresponding to the target ground state as a target parameterized quantum circuit.
9. A model processing apparatus for application to parameterized quantum circuits, comprising:
the encoding information acquisition unit is used for acquiring encoding information of a parameterized quantum circuit sample to be trained;
the calculation unit is used for calculating the second moment of the parameter gradient of the parameterized quantum circuit sample to be trained;
the sample data processing unit is used for taking the second moment of the parameter gradient of the parameterized quantum circuit sample as a label and marking the coding information of the parameterized quantum circuit sample to obtain marked coding information;
and the model training unit is used for inputting the marked coding information into a preset neural network for model training, and obtaining a target neural network after the training is finished, wherein the target neural network can predict and obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted based on the input coding information of the parameterized quantum circuit to be predicted.
10. The apparatus of claim 9, wherein the encoding information obtaining unit is further configured to determine circuit characteristics of the parameterized quantum circuit samples; and based on the circuit characteristics of the parameterized quantum circuit samples, carrying out coding processing on the parameterized quantum circuit samples to obtain the coding information of the parameterized quantum circuit samples.
11. The apparatus of claim 10, wherein the encoded information obtaining unit is further configured to determine a position of a target quantum gate in the parameterized quantum circuit sample; and at least using the position of the target quantum gate in the parameterized quantum circuit sample as the circuit characteristic of the parameterized quantum circuit sample.
12. The apparatus of claim 11, wherein the parameterized quantum circuit sample further comprises other quantum gates in addition to the target quantum gate, the other quantum gates being invariant in position within the parameterized quantum circuit sample.
13. The apparatus of any of claims 9 to 11, further comprising:
the training data acquisition unit is used for repeatedly setting the coding information according to a preset mode after the coding information of the parameterized quantum circuit sample is determined, and repeating the preset times to obtain new coding information; and obtaining a new parameterized quantum circuit based on the new coding information so as to take the new parameterized quantum circuit as a parameterized quantum circuit sample to be trained.
14. The apparatus of claim 9, wherein the model training unit is further configured to determine a loss function, wherein the loss function is determined based on a Hamiltonian of a preset physical system and a parameterized quantum circuit sample acting on the preset physical system; minimizing the loss function by adjusting parameters in the parameterized quantum circuit samples to obtain the target neural network.
15. The apparatus of claim 9 or 14, further comprising:
the model prediction unit is used for acquiring a parameterized quantum circuit to be predicted; inputting the parameterized quantum circuit to be predicted to the target neural network to obtain a second moment estimation value of the parameter gradient of the parameterized quantum circuit to be predicted;
the base state processing unit is used for selecting the parameterized quantum circuit with the second moment estimated value of the parameter gradient larger than a preset value from the parameterized quantum circuit to be predicted as an initial parameterized quantum circuit; determining a ground state of a predetermined physical system using at least the initial parameterized quantum circuit.
16. The apparatus of claim 15, wherein the ground state processing unit is further configured to select a target ground state satisfying a preset accuracy requirement from the obtained ground states of the preset physical system; and taking the initial parameterized quantum circuit corresponding to the target ground state as a target parameterized quantum circuit.
17. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-8.
18. A non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method of any one of claims 1-8.
19. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-8.
CN202011548207.4A 2020-12-23 2020-12-23 Model processing method, device, equipment and storage medium Active CN112561069B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011548207.4A CN112561069B (en) 2020-12-23 2020-12-23 Model processing method, device, equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011548207.4A CN112561069B (en) 2020-12-23 2020-12-23 Model processing method, device, equipment and storage medium

Publications (2)

Publication Number Publication Date
CN112561069A true CN112561069A (en) 2021-03-26
CN112561069B CN112561069B (en) 2021-09-21

Family

ID=75033135

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011548207.4A Active CN112561069B (en) 2020-12-23 2020-12-23 Model processing method, device, equipment and storage medium

Country Status (1)

Country Link
CN (1) CN112561069B (en)

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113379059A (en) * 2021-06-10 2021-09-10 北京百度网讯科技有限公司 Model training method for quantum data classification and quantum data classification method
CN113379056A (en) * 2021-06-02 2021-09-10 北京百度网讯科技有限公司 Quantum state data processing method and device, electronic equipment and storage medium
CN113449778A (en) * 2021-06-10 2021-09-28 北京百度网讯科技有限公司 Model training method for quantum data classification and quantum data classification method
CN114065939A (en) * 2021-11-22 2022-02-18 北京百度网讯科技有限公司 Training method, device and equipment for quantum chip design model and storage medium
CN114219076A (en) * 2021-12-15 2022-03-22 北京百度网讯科技有限公司 Quantum neural network training method and device, electronic device and medium
CN114580643A (en) * 2022-03-18 2022-06-03 北京百度网讯科技有限公司 Determination method, model processing method, device, equipment and storage medium
CN114818970A (en) * 2022-05-17 2022-07-29 北京百度网讯科技有限公司 Classical data processing method, computing device and storage medium
CN115048901A (en) * 2022-08-16 2022-09-13 阿里巴巴达摩院(杭州)科技有限公司 Quantum layout optimization method and device and computer readable storage medium
CN115049064A (en) * 2022-06-21 2022-09-13 建信金融科技有限责任公司 Data processing method and device and server
CN115374948A (en) * 2022-08-05 2022-11-22 北京百度网讯科技有限公司 Quantum neural network training method, data processing method, device and medium
CN115632660A (en) * 2022-12-22 2023-01-20 山东海量信息技术研究院 Data compression method, device, equipment and medium
CN116383707A (en) * 2023-05-08 2023-07-04 中国工商银行股份有限公司 Malicious code detection method, device, equipment and medium
CN116523053A (en) * 2022-01-24 2023-08-01 腾讯科技（深圳）有限公司 Quantum circuit simulation method, device, apparatus, storage medium and program product

Citations (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102831596A (en) * 2012-07-05 2012-12-19 中国科学院半导体研究所 Quantum dot detecting method based on mathematical morphology
CN108734299A (en) * 2017-04-19 2018-11-02 埃森哲环球解决方案有限公司 Quantum calculation machine study module
CN109074520A (en) * 2016-04-13 2018-12-21 1Qb信息技术公司 Quantum processor and its purposes for realizing neural network
US20190188604A1 (en) * 2017-12-19 2019-06-20 International Business Machines Corporation Machine learning system for predicting optimal interruptions based on biometric data colllected using wearable devices
CN110692067A (en) * 2017-06-02 2020-01-14 谷歌有限责任公司 Quantum neural network
CN110738321A (en) * 2019-10-15 2020-01-31 北京百度网讯科技有限公司 quantum signal processing method and device
CN110929798A (en) * 2019-11-29 2020-03-27 重庆邮电大学 Image classification method and medium based on structure optimization sparse convolution neural network
CN111291640A (en) * 2020-01-20 2020-06-16 北京百度网讯科技有限公司 Method and apparatus for recognizing gait
CN111368197A (en) * 2020-03-04 2020-07-03 哈尔滨理工大学 Deep learning-based comment recommendation system and method
CN111368920A (en) * 2020-03-05 2020-07-03 中南大学 Quantum twin neural network-based binary classification method and face recognition method thereof
CN111461334A (en) * 2020-03-30 2020-07-28 北京百度网讯科技有限公司 Quantum circuit processing method, device and equipment
CN111460528A (en) * 2020-04-01 2020-07-28 支付宝(杭州)信息技术有限公司 Multi-party combined training method and system based on Adam optimization algorithm
CN111563186A (en) * 2020-04-30 2020-08-21 北京百度网讯科技有限公司 Quantum data storage method, quantum data reading method, quantum data storage device, quantum data reading device and computing equipment
CN111563599A (en) * 2020-04-30 2020-08-21 合肥本源量子计算科技有限责任公司 Quantum line decomposition method and device, storage medium and electronic device
CN111630531A (en) * 2018-01-18 2020-09-04 谷歌有限责任公司 Classification using quantum neural networks
US20200293936A1 (en) * 2019-03-15 2020-09-17 Microsoft Technology Licensing, Llc Phase estimation with randomized hamiltonians
CN111814907A (en) * 2020-07-28 2020-10-23 南京信息工程大学 Quantum generation countermeasure network algorithm based on condition constraint
CN112001498A (en) * 2020-08-14 2020-11-27 苏州浪潮智能科技有限公司 Data identification method and device based on quantum computer and readable storage medium

Patent Citations (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102831596A (en) * 2012-07-05 2012-12-19 中国科学院半导体研究所 Quantum dot detecting method based on mathematical morphology
CN109074520A (en) * 2016-04-13 2018-12-21 1Qb信息技术公司 Quantum processor and its purposes for realizing neural network
CN108734299A (en) * 2017-04-19 2018-11-02 埃森哲环球解决方案有限公司 Quantum calculation machine study module
CN110692067A (en) * 2017-06-02 2020-01-14 谷歌有限责任公司 Quantum neural network
US20190188604A1 (en) * 2017-12-19 2019-06-20 International Business Machines Corporation Machine learning system for predicting optimal interruptions based on biometric data colllected using wearable devices
CN111630531A (en) * 2018-01-18 2020-09-04 谷歌有限责任公司 Classification using quantum neural networks
US20200293936A1 (en) * 2019-03-15 2020-09-17 Microsoft Technology Licensing, Llc Phase estimation with randomized hamiltonians
CN110738321A (en) * 2019-10-15 2020-01-31 北京百度网讯科技有限公司 quantum signal processing method and device
CN110929798A (en) * 2019-11-29 2020-03-27 重庆邮电大学 Image classification method and medium based on structure optimization sparse convolution neural network
CN111291640A (en) * 2020-01-20 2020-06-16 北京百度网讯科技有限公司 Method and apparatus for recognizing gait
CN111368197A (en) * 2020-03-04 2020-07-03 哈尔滨理工大学 Deep learning-based comment recommendation system and method
CN111368920A (en) * 2020-03-05 2020-07-03 中南大学 Quantum twin neural network-based binary classification method and face recognition method thereof
CN111461334A (en) * 2020-03-30 2020-07-28 北京百度网讯科技有限公司 Quantum circuit processing method, device and equipment
CN111460528A (en) * 2020-04-01 2020-07-28 支付宝(杭州)信息技术有限公司 Multi-party combined training method and system based on Adam optimization algorithm
CN111563186A (en) * 2020-04-30 2020-08-21 北京百度网讯科技有限公司 Quantum data storage method, quantum data reading method, quantum data storage device, quantum data reading device and computing equipment
CN111563599A (en) * 2020-04-30 2020-08-21 合肥本源量子计算科技有限责任公司 Quantum line decomposition method and device, storage medium and electronic device
CN111814907A (en) * 2020-07-28 2020-10-23 南京信息工程大学 Quantum generation countermeasure network algorithm based on condition constraint
CN112001498A (en) * 2020-08-14 2020-11-27 苏州浪潮智能科技有限公司 Data identification method and device based on quantum computer and readable storage medium

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
ANURAG ANSHU等: "Sample-efficient learning of quantum many-body systems", 《ARXIV:2004.07266》 *
GUANGXI LI等: "VSQL: Variational Shadow Quantum Learning for Classification", 《ARXIV:2012.08288V1》 *
JARROD R. MCCLEAN等: "Barren plateaus in quantum neural network training landscapes", 《ARXIV:1803.11173V1》 *
孙晓明: "量子计算若干前沿问题综述", 《中国科学:信息科学》 *
陈小余等: "量子高斯态的纠缠原因分析", 《第十一届全国量子光学学术会议论文摘要集》 *

Cited By (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113379056A (en) * 2021-06-02 2021-09-10 北京百度网讯科技有限公司 Quantum state data processing method and device, electronic equipment and storage medium
CN113379056B (en) * 2021-06-02 2023-10-31 北京百度网讯科技有限公司 Quantum state data processing method and device, electronic equipment and storage medium
CN113379059B (en) * 2021-06-10 2022-09-23 北京百度网讯科技有限公司 Model training method for quantum data classification and quantum data classification method
CN113449778A (en) * 2021-06-10 2021-09-28 北京百度网讯科技有限公司 Model training method for quantum data classification and quantum data classification method
CN113449778B (en) * 2021-06-10 2023-04-21 北京百度网讯科技有限公司 Model training method for quantum data classification and quantum data classification method
CN113379059A (en) * 2021-06-10 2021-09-10 北京百度网讯科技有限公司 Model training method for quantum data classification and quantum data classification method
CN114065939A (en) * 2021-11-22 2022-02-18 北京百度网讯科技有限公司 Training method, device and equipment for quantum chip design model and storage medium
CN114219076A (en) * 2021-12-15 2022-03-22 北京百度网讯科技有限公司 Quantum neural network training method and device, electronic device and medium
CN116523053B (en) * 2022-01-24 2024-09-10 腾讯科技（深圳）有限公司 Quantum circuit simulation method, device, apparatus, storage medium and program product
CN116523053A (en) * 2022-01-24 2023-08-01 腾讯科技（深圳）有限公司 Quantum circuit simulation method, device, apparatus, storage medium and program product
CN114580643A (en) * 2022-03-18 2022-06-03 北京百度网讯科技有限公司 Determination method, model processing method, device, equipment and storage medium
CN114580643B (en) * 2022-03-18 2023-04-28 北京百度网讯科技有限公司 Determination method, model processing method, device, equipment and storage medium
CN114818970A (en) * 2022-05-17 2022-07-29 北京百度网讯科技有限公司 Classical data processing method, computing device and storage medium
CN115049064A (en) * 2022-06-21 2022-09-13 建信金融科技有限责任公司 Data processing method and device and server
CN115374948A (en) * 2022-08-05 2022-11-22 北京百度网讯科技有限公司 Quantum neural network training method, data processing method, device and medium
CN115374948B (en) * 2022-08-05 2024-07-26 北京百度网讯科技有限公司 Training method, data processing method, device and medium of quantum neural network
CN115048901A (en) * 2022-08-16 2022-09-13 阿里巴巴达摩院(杭州)科技有限公司 Quantum layout optimization method and device and computer readable storage medium
CN115632660B (en) * 2022-12-22 2023-03-17 山东海量信息技术研究院 Data compression method, device, equipment and medium
CN115632660A (en) * 2022-12-22 2023-01-20 山东海量信息技术研究院 Data compression method, device, equipment and medium
CN116383707A (en) * 2023-05-08 2023-07-04 中国工商银行股份有限公司 Malicious code detection method, device, equipment and medium

Also Published As

Publication number Publication date
CN112561069B (en) 2021-09-21

Similar Documents

Publication Publication Date Title
CN112561069B (en) Model processing method, device, equipment and storage medium
CN109657805B (en) Hyper-parameter determination method, device, electronic equipment and computer readable medium
CN114219076A (en) Quantum neural network training method and device, electronic device and medium
CN113705793B (en) Decision variable determination method and device, electronic equipment and medium
CN115374948B (en) Training method, data processing method, device and medium of quantum neural network
CN114021728B (en) Quantum data measuring method and system, electronic device, and medium
CN112529024B (en) Sample data generation method and device and computer readable storage medium
CN114358319B (en) Machine learning framework-based classification method and related device
CN112966744A (en) Model training method, image processing method, device and electronic equipment
CN114580649A (en) Method and device for eliminating quantum Pagli noise, electronic equipment and medium
CN114240555A (en) Click rate prediction model training method and device and click rate prediction method and device
CN115577776A (en) Method, device and equipment for determining ground state energy and storage medium
CN113449778A (en) Model training method for quantum data classification and quantum data classification method
CN115577787A (en) Quantum amplitude estimation method, device, equipment and storage medium
CN113010687B (en) Exercise label prediction method and device, storage medium and computer equipment
CN113255824A (en) Method and device for training classification model and data classification
CN115018078B (en) Quantum circuit operation method and device, electronic equipment and medium
Balakrishnan et al. Quantum Neural Network for Time Series Forecasting: Harnessing Quantum Computing's Potential in Predictive Modeling
CN116306952B (en) Molecular property prediction method and device, storage medium and electronic device
CN116341670B (en) Method and device for processing graph network data through quantum node embedding algorithm
CN113420561B (en) Named entity identification method, device, equipment and storage medium
CN116090571A (en) Quantum linear solving method, device and medium based on generalized minimum residual quantity
CN117875370A (en) Task processing method and device utilizing molecular data
CN115482111A (en) QSVM-based multi-factor stock selection method and related device
CN117669751A (en) Quantum circuit simulation method and device and electronic equipment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant