Image classification method and system based on novel quantum convolutional network

Abstract

According to the image classification method and system based on the novel quantum convolutional network, complex dimension reduction operation does not need to be carried out on an image to be classified, data rearrangement is carried out twice according to a convolution kernel sliding mode and a binary inversion mode, and then the data are coded to the amplitude of a quantum state; compared with the existing quantum convolutional neural network for image classification, the consumption of computing resources is reduced to a great extent; in addition, information such as edges and positions can be reserved through rearrangement operation and amplitude coding operation of the to-be-classified images, convolution operation is carried out through the parameterized quantum lines, and the effect of extracting local features can be achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of image classification, and in particular relates to an image classification method and system based on a novel quantum convolutional network. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In the era of big data, image data is growing at an alarming rate, and contains rich visual information. Classifying and analyzing these massive amounts of image data can help people dig out valuable information from them. Image classification can provide more comprehensive and accurate data support for decision-making, thereby promoting the development and progress of various fields.

[0004] Deep learning methods have shown excellent results in image classification problems, among which convolutional neural network models are particularly outstanding. As an important branch in the field of artificial intelligence, deep learning has made significant progress in the past decade. It has made great progress in algorithms, computing power and data, making artificial intelligence technology break through the limitations of commercial use, and the industry application scenarios are becoming more and more extensive, and the scale of the industry continues to expand. As a commonly used network model in the field of deep learning, the classic convolutional neural network has important performance in image classification problems. Among them, the local feature extraction ability is crucial for the classic convolutional neural network (CNN), which can help the classic convolutional neural network model better capture local information, reduce the amount of calculation, and promote network depth. However, the classic convolutional neural network has some limitations. Because it contains a large number of parameters, it not only increases the complexity of the model, but also makes the training process time-consuming and requires a lot of computing resources. And as the size of image data increases, these limitations will further limit the effect of convolutional neural networks on image classification problems.

[0005] With the potential for quantum computers to outperform any foreseeable classical computers in solving certain computing tasks, quantum deep learning has become a potential solution to the challenge of processing increasing amounts of data. Although some quantum convolutional neural networks for image classification have been proposed, existing quantum convolutional neural networks do not have the ability to extract local features, and these quantum convolutional neural networks need to first use PCA and other methods to reduce the dimensionality of classical data, which still takes up a lot of classical computing resources. Moreover, as the number of classification categories increases, the existing quantum convolutional neural networks are less effective.

[0006] Therefore, while using quantum computing to solve the problem of classical convolutional neural networks, which consumes a lot of computing resources, it is a technical problem that needs to be solved at present to effectively retain the important ability of classical convolutional neural networks, which is local feature extraction, so that it can show better generalization ability in image classification problems. Summary of the invention

[0007] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides an image classification method and system based on a novel quantum convolutional network, which performs two data rearrangements according to the convolution kernel sliding method and the binary inversion method, and then encodes them into the amplitude of the quantum state, which greatly reduces the consumption of computing resources. Then, the convolution operation is performed using a parameterized quantum circuit, which can achieve the effect of extracting local features.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] In a first aspect, the present invention provides an image classification method based on a novel quantum convolutional network, comprising:

[0010] Acquire an image to be classified, and convert the acquired image to be classified into a one-dimensional data feature;

[0011] The converted one-dimensional data features are rearranged in turn according to the convolution kernel sliding method and the binary inversion method;

[0012] Encode the rearranged one-dimensional data features into the amplitude of the quantum bit;

[0013] Convolution of the input quantum state of the qubit using parameterized quantum circuits;

[0014] A quantum measurement operation is performed on the result of the convolution processing to obtain the corresponding eigenvalue, and the image classification result is obtained based on the obtained eigenvalue.

[0015] In a second aspect, the present invention provides an image classification system based on a novel quantum convolutional network, comprising:

[0016] A conversion module is configured to: obtain an image to be classified, and convert the obtained image to be classified into a one-dimensional data feature;

[0017] A rearrangement module is configured to: rearrange the converted one-dimensional data features in sequence according to a convolution kernel sliding mode and a binary inversion mode;

[0018] An encoding module, which is configured to: encode the rearranged one-dimensional data feature into the amplitude of the quantum bit;

[0019] A convolution module, configured to: perform convolution processing on an input quantum state of a quantum bit using a parameterized quantum circuit;

[0020] The quantum measurement module is configured to: perform quantum measurement operation on the result of convolution processing to obtain corresponding eigenvalues, and obtain image classification results based on the obtained eigenvalues.

[0021] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in the first aspect is performed.

[0022] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method described in the first aspect is performed.

[0023] One or more of the above technical solutions have the following beneficial effects:

[0024] In the present invention, there is no need to perform complex dimensionality reduction operations on the image to be classified. It is only necessary to rearrange the data twice according to the convolution kernel sliding method and the binary inversion method, and then encode it into the amplitude of the quantum state. Compared with the existing quantum convolutional neural network for image classification, the consumption of computing resources is greatly reduced; and the rearrangement operation and amplitude encoding operation of the image to be classified can retain information such as edges and positions, and the convolution operation using parameterized quantum circuits can achieve the effect of extracting local features.

[0025] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0027] Figure 1 This is a flow chart of a quantum convolutional neural network with local feature extraction capability in Embodiment 1 of the present invention;

[0028] Figure 2 It is an existing quantum convolutional neural network for image binary classification;

[0029] Figure 3 The quantum convolutional neural network with local feature extraction capability in the first embodiment of the present invention;

[0030] Figure 4 It is a partial matrix of the parameterized circuit Ansatz of the quantum convolutional neural network with local feature extraction capability of the present invention in the first embodiment of the present invention;

[0031] Figure 5 The loss function of the quantum convolutional neural network with local feature extraction capability for processing image binary classification in the first embodiment of the present invention;

[0032] Figure 6 The loss function of the quantum convolutional neural network with local feature extraction capability for processing four-class image classification in the first embodiment of the present invention;

[0033] Figure 7 The loss function of the quantum convolutional neural network with local feature extraction capability for processing seven image classifications in the first embodiment of the present invention;

[0034] Figure 8 Schematic diagram of the quantum bit route designed in Example 1 of the present invention. DETAILED DESCRIPTION

[0035] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0036] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0037] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0038] Embodiment 1

[0039] Local feature extraction is a concept in image processing and computer vision, which refers to the process of extracting information and features from local areas of an image (such as small pixel blocks or superpixels). These features usually include attributes such as edges, textures, colors, and shapes, which can describe the details and structures in the image.

[0040] Using local feature extraction can extract key information from raw image data, reduce the amount of data, and facilitate subsequent processing. This method can adapt to different image content and scenes, provide support for various visual tasks, and is the basis for many innovative applications. The extracted local features can form feature representations at different levels, from simple edges and textures to complex shapes and structures, and can have good robustness to image noise and occlusion. Even if some parts of the image are damaged or occluded, the main objects in the image can still be identified. It is the basis of pattern recognition and image classification, and can provide machine learning algorithms with the information needed to identify and classify image content. In convolutional neural networks (CNNs), local feature extraction is achieved through convolutional layers, which use learnable convolution kernels to automatically extract local features of images. This is one of the key factors for the success of CNNs in visual tasks and can greatly improve computational efficiency.

[0041] Convolutional Neural Networks (CNN) is a deep learning model that performs well in image recognition and processing tasks, mainly due to its powerful local feature extraction capabilities. CNN automatically extracts local features through convolutional layers, which use small convolution kernels to slide over the input image to capture key information in local areas, such as edges, textures, and shapes. The importance of local feature extraction lies in its ability to extract meaningful patterns from complex image data, which are the basis for image recognition and classification. Since the parameters of the convolution kernel are shared throughout the network, CNN is able to learn rich feature representations with fewer parameters, which not only improves computational efficiency but also enhances the model's ability to generalize to new data.

[0042] The hierarchical structure of CNN allows the model to gradually learn features from simple to complex. The bottom layer may focus on basic edge features, while the upper layer may recognize more complex shapes and object features. This hierarchical feature extraction is the key to the success of CNN in visual tasks and the reason why it is widely used in many fields such as image classification, object detection and image segmentation.

[0043] Quantum Neural Networks (QNN) are cutting-edge technologies that combine quantum computing and neural networks. They use quantum bits (qubits) to encode information and apply quantum gates to perform weight matrix operations similar to those in traditional neural networks. The core advantage of QNNs is that they can use the properties of quantum mechanics, such as superposition and entanglement, to handle complex computing tasks. Among them, Quantum Convolutional Neural Networks (QCNN) have shown unique potential in the field of image classification. Compared with classical CNNs, QCNNs are able to perform efficient feature extraction in quantum state space, which gives them potential computational advantages when processing high-dimensional data.

[0044] Although quantum convolutional neural networks have great potential in theory, they are still in the early stages of research and development, and the limitations of quantum hardware are the main challenges currently faced. However, with the continuous advancement of quantum computing technology, quantum convolutional neural networks are expected to play an important role in future image processing and classification tasks.

[0045] This embodiment discloses an image classification method based on a novel quantum convolutional network, comprising:

[0046] Acquire an image to be classified, and convert the acquired image to be classified into a one-dimensional data feature;

[0047] The converted one-dimensional data features are rearranged in turn according to the convolution kernel sliding method and the binary inversion method;

[0048] Encode the rearranged one-dimensional data features into the amplitude of the quantum bit;

[0049] Convolution of the input quantum state of the qubit using parameterized quantum circuits;

[0050] A quantum measurement operation is performed on the result of the convolution processing to obtain the corresponding eigenvalue, and the image classification result is obtained based on the obtained eigenvalue.

[0051] This embodiment draws on the classic convolutional neural network, emphasizes the role of local feature extraction capability for the classic convolutional neural network, refers to the important method of local feature extraction and implements it through data preprocessing and designing parameterized circuit Ansatz. Figure 1 The main process of this model can be simply divided into two parts: data preprocessing and training the model to obtain prediction results.

[0052] The following is a detailed introduction to these two parts.

[0053] Step 1: Data Preprocessing

[0054] As shown in Table 1, this embodiment uses the data of an image in the MNIST dataset as an example to introduce in detail:

[0055] Step 11: Expand the image data from 28*28 to 32*32 by zero padding the edges, which is conducive to the data encoding to be more consistent with the logical wave function; then flatten the image data into 1024-dimensional one-dimensional data, and obtain the initial two rows in Table 1, where indexes 0 to 1023 are image data a0 to a 1023 .

[0056] Step 12: First image data reordering.

[0057] According to the classic convolutional neural network, the 4*4 convolution kernel is slid with a step size of 4, and the 16 data of the image data are convolved each time as a group and arranged into 64 groups of data. For example, the first group of 16 data is a0, a1, a2, a3, a 32 、a 33 、a 34 、a 35 、a 64 、a 65 、a 66 、a 67 、a 96 、a 97 、a 98 、a 99 , these 16 data are arranged in order at the first 16 positions, the convolution kernel slides to the right with a step size of 4, and contains 16 data again and arranges them at positions 17-32, and so on, to obtain the two rows in Table 1 after one sorting. It can be seen from Table 1 that after one sorting, the image data at some indexes have changed, such as the data at 96 has changed from a 96 becomes a 24 , the data at 97 is from a 97 becomes a 25 , while the data at 512, as the last data of the 32nd group, has not changed.

[0058] Step 13: Second data reordering.

[0059] This is mainly done for the programming framework used in this embodiment. Since the high and low positions of the programming framework are opposite, the binary string corresponding to the index needs to be reversed. For example, the binary string at position 1 is 0000000001, and after reversal, it is 1000000000, which corresponds to 512. Therefore, the data at position 1 needs to be swapped with the data at position 512. After this operation, the two rows in the secondary sorting in Table 1 are obtained.

[0060] At this point, the data preprocessing part is completed. Through the above steps, the MNIST data set is processed into the data required by the model of the present invention, which is convenient for subsequent operations.

[0061] Table 1:

[0062]

[0063]

[0064] Step 2: Model training and testing.

[0065] The following describes the composition of the quantum convolutional neural network model with local feature extraction capability of the present invention:

[0066] 1. Data encoding.

[0067] Data encoding is the effective mapping of classical data to quantum states so that it can be processed using the advantages of quantum computing. Common quantum data encoding methods include direct encoding, angle encoding, amplitude encoding, etc. This embodiment uses amplitude encoding to encode 1024-dimensional image data into the amplitude of 10 quantum bits, which is also a major advantage of quantum computing.

[0068] 2. Design of parameterized quantum circuit Ansatz.

[0069] like Figure 2 As shown in the figure, the quantum convolutional neural network is an existing network model for binary image classification problems. Excluding the data encoding and measurement parts, the remaining part is the parameterized quantum circuit. It can be seen that the parameterized quantum circuit of this model is relatively complex.

[0070] like Figure 8 As shown, the parameterized quantum circuit of this embodiment uses a four-qubit circuit with an All-to-All structure, and adds R X and R Z After two quantum gates with parameters, each qubit is used as a control bit to exert a controlled R on the remaining three qubits. X Operation to achieve full connection.

[0071] The network model of this embodiment is as follows: Figure 3 As shown in Figure 1, since the data is reordered, only one U operation with 20 parameters is needed to perform a convolution process on the data.

[0072] like Figure 4The figure shows the matrix representation of the model parameterized quantum circuit. The matrix size is 1024*1024, with only 16*16 parameter-related values ​​on the diagonal, and the rest of the positions are zero. Each row of this 16*16 matrix is ​​equivalent to a 4*4 convolution kernel, which is multiplied with 16 image data for convolution. The 16 rows are 16 different convolution kernels.

[0073] After processing the designed matrix, the image data is equivalent to a classic convolution process with a convolution kernel size of 4*4, a step size of 4, and 16 channels.

[0074] 3. Quantum measurement.

[0075] In order to obtain the eigenvalues ​​extracted after convolution of the image data, the model uses a probability measurement method to obtain the square of 1024 amplitudes of 10 quantum bits. The measured values ​​are the eigenvalues ​​after convolution of 16 convolution kernels, 64 eigenvalues ​​per channel, and the eigenvalues ​​of a single channel or multiple channels can be selected as the basis for classification and post-processing.

[0076] After the model is built, it can be trained and tested using the processed image data. In this example, the initial learning rate of the model is set to 0.1, the optimizer uses the Adam optimizer, the loss function uses the cross entropy loss CrossEntropyLoss, and the test is performed after training for 10 epochs. The specific results will be shown later.

[0077] The quantum convolutional neural network for image classification is one of the important research directions combining quantum computing and deep learning. Compared with other quantum convolutional neural networks for image classification, this embodiment has the following advantages:

[0078] (1) There is no need to perform complex dimensionality reduction on the data, only two simple reorderings are required. The preprocessed data is encoded into the amplitude of the quantum state by amplitude coding. Compared with other existing quantum convolutional neural networks for image classification, this embodiment greatly reduces the consumption of computing resources.

[0079] (2) Possess the ability to extract local features. Zero pad the edges of the processed data, encode the image data sequentially into the amplitude of each quantum state through amplitude coding, retain edge information, position information, etc., and design the parameterized quantum circuit Ansatz to realize local feature extraction.

[0080] (3) Lower model complexity. Existing quantum convolutional neural networks require multiple convolutional blocks, such as Figure 2 , while the parameterized quantum circuit of this embodiment only needs one convolution block, such as Figure 3 shown.

[0081] (4) The model has stronger generalization ability. Existing quantum convolutional neural network models only show good results for the binary or quadratic classification of the MNIST dataset, but perform poorly for more classification categories. This model shows good classification results for the binary, quadratic, and heptad classification of the MNIST dataset.

[0082] In this example, the MNIST data set is used to test the quantum convolutional neural network model, and the accuracy of the model in two-category, four-category, and seven-category problems is tested respectively. Compared with the existing quantum convolutional neural network (two-category accuracy 98.4%, four-category accuracy 90.03%, seven-category accuracy about 50%), the model of this embodiment shows better results with the increase of image classification categories, and the complexity of the circuit is lower. The specific results are as follows:

[0083] In the problem of binary image classification, this embodiment performs very well, with a final accuracy rate of 99.80%, and the loss function decreases rapidly and converges quickly. Figure 5 The prediction results of the model are shown in Table 2.

[0084] Table 2: Binary classification prediction

[0085]

[0086] The effect of this embodiment on the four-class image classification is still good, with an accuracy rate of 92.74%. The loss function decreases smoothly and gradually reaches a convergence state. Figure 6 The final prediction of the four-category model image is shown in Table 3.

[0087] Table 3: Four-category prediction

[0088]

[0089] In the classification problems of more categories where other quantum convolutional neural networks perform poorly, the performance of the present invention is relatively better, such as Figure 7 As shown in Table 4, the classification accuracy of the seven categories can reach 86.69%, and the loss function decreases smoothly, but the fluctuation is relatively large after convergence. The specific prediction results are shown in Table 4.

[0090] Table 4: Seven-category predictions

[0091]

[0092] Embodiment 2

[0093] The purpose of this embodiment is to provide an image classification system based on a novel quantum convolutional network, including:

[0094] A conversion module is configured to: obtain an image to be classified, and convert the obtained image to be classified into a one-dimensional data feature;

[0095] A rearrangement module is configured to: rearrange the converted one-dimensional data features in sequence according to a convolution kernel sliding mode and a binary inversion mode;

[0096] An encoding module, which is configured to: encode the rearranged one-dimensional data feature into the amplitude of the quantum bit;

[0097] A convolution module, configured to: perform convolution processing on an input quantum state of a quantum bit using a parameterized quantum circuit;

[0098] The quantum measurement module is configured to: perform quantum measurement operation on the result of convolution processing to obtain corresponding eigenvalues, and obtain image classification results based on the obtained eigenvalues.

[0099] In further embodiments, there is also provided:

[0100] An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in Embodiment 1 is performed. For the sake of brevity, no further description is given here.

[0101] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0102] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0103] A computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the method described in embodiment 1 is completed.

[0104] The method in the first embodiment can be directly embodied as a hardware processor, or a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware. To avoid repetition, it will not be described in detail here.

[0105] A computer program product includes a computer program, and when the computer program is executed by a processor, the method described in the first embodiment is implemented.

[0106] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer executable instructions, such as instructions included in a program module, which are executed in a device on a real or virtual processor of the target to perform the process / method as described above. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules can be combined or divided between program modules as needed. Machine executable instructions for program modules can be executed in local or distributed devices. In distributed devices, program modules can be located in local and remote storage media.

[0107] The computer program code for implementing the method of the present invention can be written in one or more programming languages. These computer program codes can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the computer or other programmable data processing device, causes the function / operation specified in the flow chart and / or block diagram to be implemented. The program code can be executed completely on a computer, partially on a computer, as an independent software package, partially on a computer and partially on a remote computer or completely on a remote computer or server.

[0108] In the context of the present invention, computer program codes or related data may be carried by any appropriate carrier to enable a device, apparatus or processor to perform the various processes and operations described above. Examples of carriers include signals, computer readable media, etc. Examples of signals may include electrical, optical, radio, acoustic or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0109] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software 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.

[0110] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. An image classification method based on a novel quantum convolutional network, characterized in that: include: Acquire an image to be classified, and convert the acquired image to be classified into a one-dimensional data feature; The converted one-dimensional data features are rearranged in turn according to the convolution kernel sliding method and the binary inversion method; Encode the rearranged one-dimensional data features into the amplitude of the quantum bit; Convolution of the input quantum state of the qubit using parameterized quantum circuits; A quantum measurement operation is performed on the result of the convolution processing to obtain the corresponding eigenvalue, and the image classification result is obtained based on the obtained eigenvalue.

2. The image classification method based on a novel quantum convolutional network as claimed in claim 1, characterized in that: Before converting the acquired image to be classified into a one-dimensional data feature, it also includes performing an expansion operation on the image to be classified so that its data encoding conforms to the logical wave function.

3. The image classification method based on a novel quantum convolutional network as claimed in claim 1, characterized in that: The converted one-dimensional data features are rearranged in sequence according to the convolution kernel sliding method, specifically: Determine the convolution kernel size and sliding step size; The data subjected to convolution each time is regarded as a group of data, and the indexes are arranged in sequence to obtain the one-dimensional data features after the first rearrangement.

4. The image classification method based on a novel quantum convolutional network as claimed in claim 1, characterized in that: The one-dimensional data features rearranged in a sliding manner according to the convolution kernel are rearranged by a binary inversion method. Specifically, the binary strings corresponding to the indexes after the first rearrangement are reversed in sequence, and the data features under the indexes corresponding to the reversed binary strings are interchanged with the data features under the indexes before the inversion.

5. An image classification method based on a novel quantum convolutional network as described in any one of claims 1 to 4, characterized in that: It also includes maximum and minimum normalization processing of one-dimensional data features after rearrangement by convolution kernel sliding method and binary inversion method.

6. The image classification method based on a novel quantum convolutional network as claimed in claim 1, characterized in that: Amplitude coding is used to encode the rearranged one-dimensional data features into the amplitude of the quantum bit.

7. The image classification method based on a novel quantum convolutional network as claimed in claim 1, characterized in that: The parameterized quantum circuit uses a four-qubit circuit with an All-to-All structure, and adds a parameterized quantum gate R to each of the four qubits. X and R Z After that, each qubit is used as a control bit to apply a controlled quantum gate R to the remaining three qubits. X Operation to achieve full connection.

8. An image classification system based on a novel quantum convolutional network, characterized in that: include: A conversion module is configured to: obtain an image to be classified, and convert the obtained image to be classified into a one-dimensional data feature; A rearrangement module is configured to: rearrange the converted one-dimensional data features in sequence according to a convolution kernel sliding mode and a binary inversion mode; An encoding module, which is configured to: encode the rearranged one-dimensional data feature into the amplitude of the quantum bit; A convolution module, configured to: perform convolution processing on an input quantum state of a quantum bit using a parameterized quantum circuit; The quantum measurement module is configured to: perform quantum measurement operation on the result of convolution processing to obtain corresponding eigenvalues, and obtain image classification results based on the obtained eigenvalues.

9. An electronic device, characterized in that: The method comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 7 is completed.

10. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the method described in any one of claims 1 to 7.

Images