A bank fixed deposit target customer classification method based on a quantum neural network
By processing bank time deposit data using quantum neural networks, the computational complexity and large data volume issues of classical machine learning in classifying bank time deposit data are solved, achieving more efficient customer classification, improving the success rate of marketing campaigns and reducing costs.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHONGKE YUNCHAO (BEIJING) QUANTUM TECH CO LTD
- Filing Date
- 2024-03-27
- Publication Date
- 2026-05-29
AI Technical Summary
In existing technologies, classic machine learning suffers from computational complexity, large training data volume, and long time requirements in classifying bank time deposit data, making it difficult to effectively improve the success rate of marketing campaigns.
A quantum neural network-based method is used to preprocess bank time deposit data, loading customer feature information into quantum states, processing it using a quantum neural network, and updating parameters through a loss function until the termination condition is met, thereby achieving customer classification.
Quantum neural networks, through the parallelism and non-locality of quantum computing, improve the speed of customer information feature extraction and convergence, reduce the amount of training data and model design complexity, are applicable to existing NISQ real quantum computers, and improve the success rate of time deposit marketing while reducing costs.
Smart Images
Figure CN118245875B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of quantum computing-based classification technology, and more particularly to a method for classifying target customers of bank time deposits based on quantum neural networks. Background Technology
[0002] Term deposits are a major source of revenue for banks. Term deposits are cash investments held by financial institutions. Your money is invested at a predetermined interest rate for a fixed period or term. To market term deposits to customers, banks employ various promotional programs, including email marketing, advertising, telemarketing, and digital marketing. Telemarketing remains one of the most effective ways to reach people. However, it requires significant investment, as banks need to employ large call centers to actually execute the campaigns to market term deposits. Therefore, identifying the most likely customers to convert beforehand is crucial. This allows for targeted calls, avoiding the indiscriminate use of large call centers to inquire about term deposit intentions. Targeted marketing with a broad audience significantly increases the success rate of marketing campaigns. Currently, classical machine learning is primarily used to process bank term deposit data, while the classification of target customers using quantum computing-based data is yet to be explored.
[0003] Currently, classical machine learning is primarily used to process bank time deposit data. Classical computers perform calculations based on classical bits, which differs significantly from quantum computers, which process information using qubits. With the same number of bits, quantum computers can process far more data due to their vast Hilbert space. Using classical neural networks to learn from time deposit data requires complex neural network model design and hyperparameter tuning techniques, and demands a large amount of data and significant time for training. Leveraging the characteristics of quantum computing, quantum neural networks can achieve results in classifying target customers for bank time deposits that are unattainable by purely classical neural network models. Summary of the Invention
[0004] The technical problem to be solved by this invention is to address the shortcomings of existing technologies, specifically by providing a method for classifying target customers for bank time deposits based on quantum neural networks, as detailed below:
[0005] 1) In a first aspect, the present invention provides a method for classifying target customers of bank time deposits based on quantum neural networks, the specific technical solution of which is as follows:
[0006] Preprocess the bank time deposit dataset to obtain the feature information of bank time deposit customers;
[0007] Load the characteristic information of bank time deposit customers into a quantum state;
[0008] A quantum neural network is used to process quantum states to obtain prediction results of the characteristic information of bank time deposit customers. A loss function is constructed based on the prediction results of the characteristic information of bank time deposit customers and the real labels of bank time deposit customers. The parameters of the quantum neural network are updated according to the loss function until the termination condition is met.
[0009] The latest quantum neural network is used to classify the characteristic information of the preset users' bank time deposit customers.
[0010] The beneficial effects of the quantum neural network-based target customer classification method for bank time deposits provided by this invention are as follows:
[0011] The quantum neural network proposed in this invention employs a novel computing paradigm based on the fundamental principles of quantum mechanics—quantum computing. Leveraging the entanglement and superposition characteristics of quantum computing physical systems, it can learn pattern information from data that classical computers cannot acquire. Furthermore, the powerful parallelism and non-locality of quantum neural networks outperform classical neural networks in extracting global features and convergence speed from bank deposit customer information. This results in quantum neural networks requiring less data for training compared to classical neural networks, and their design and hyperparameter tuning are simpler. Moreover, this invention can be implemented using existing NISQ quantum computers.
[0012] Based on the above scheme, the present invention can be further improved as follows: a method for classifying target customers of bank time deposits based on quantum neural networks.
[0013] Furthermore, the characteristic information of bank time deposit customers is loaded into a quantum state, including:
[0014] The number of qubits to be used is determined based on the characteristics of bank time deposit customers.
[0015] Based on the number of qubits used and using angle encoding, the characteristic information of bank time deposit customers is loaded into quantum states.
[0016] Furthermore, by processing the quantum state using a quantum neural network, prediction results of characteristic information of bank time deposit customers are obtained, including:
[0017] The quantum state is processed using a quantum neural network, the nth qubit of the quantum neural network is measured, and the prediction result of the characteristic information of the bank's time deposit customer is obtained based on the measurement result of the nth qubit.
[0018] Furthermore, the parameters of the quantum neural network are updated according to the loss function, including: updating the parameters of the quantum neural network according to the loss function, and updating the parameters of the quantum neural network using the parameter shifting rule.
[0019] 2) Secondly, the present invention also provides a target customer classification system for bank time deposits based on quantum neural networks, the specific technical solution of which is as follows:
[0020] It includes a data preprocessing module, a quantum state loading module, an update module, and a feature information classification module;
[0021] The data preprocessing module is used to preprocess the bank's time deposit dataset to obtain the characteristic information of bank time deposit customers;
[0022] The quantum state loading module is used to: load the characteristic information of bank time deposit customers into quantum states;
[0023] The update module is used to: process quantum states using a quantum neural network to obtain prediction results of the feature information of bank time deposit customers; construct a loss function based on the prediction results of the feature information of bank time deposit customers and the real labels of bank time deposit customers; and update the parameters of the quantum neural network based on the loss function until the termination condition is met.
[0024] The feature information classification module is used to classify the feature information of the preset users' bank time deposit customers based on the latest quantum neural network.
[0025] Based on the above scheme, the target customer classification system for bank time deposits based on quantum neural networks of the present invention can be further improved as follows.
[0026] The quantum state loading module is specifically used for:
[0027] The number of qubits to be used is determined based on the characteristics of bank time deposit customers.
[0028] Based on the number of qubits used and using angle encoding, the characteristic information of bank time deposit customers is loaded into quantum states.
[0029] Furthermore, the update module is specifically used to: process the quantum state using a quantum neural network, measure the nth qubit of the quantum neural network, and obtain the prediction result of the characteristic information of the bank's time deposit customer based on the measurement result of the nth qubit.
[0030] Furthermore, the update module is specifically used to: update the parameters of the quantum neural network according to the loss function, and update the parameters of the quantum neural network using the parameter shifting rule.
[0031] 3) In a third aspect, the present invention also provides a computer device, the computer device including a processor coupled to a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor, so that the computer device implements any of the above-mentioned methods for classifying target customers of bank time deposits based on quantum neural networks.
[0032] 4) In a fourth aspect, the present invention also provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to enable a computer to implement any of the above-mentioned methods for classifying target customers of bank time deposits based on quantum neural networks.
[0033] It should be noted that the beneficial effects of the technical solutions of the second to fourth aspects of the present invention and their corresponding possible implementations can be found in the above description of the technical effects of the first aspect and its corresponding possible implementations, and will not be repeated here. Attached Figure Description
[0034] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:
[0035] Figure 1 This is a flowchart illustrating a method for classifying target customers for bank time deposits based on a quantum neural network, according to an embodiment of the present invention.
[0036] Figure 2 A schematic diagram of a quantum circuit;
[0037] Figure 3 The network structure of a quantum neural network;
[0038] Figure 4 For a U-shaped network structure;
[0039] Figure 5 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. Detailed Implementation
[0040] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.
[0041] like Figure 1 As shown in the figure, an embodiment of the present invention provides a method for classifying target customers for bank time deposits based on quantum neural networks, which includes the following steps:
[0042] S1. Preprocess the bank's time deposit dataset to obtain the feature information of bank time deposit customers, specifically:
[0043] Download the Banking dataset, which contains bank time deposit data. Then, represent the non-numeric categories of some features in the downloaded bank time deposit dataset using different numeric levels to obtain the processed 16-dimensional feature information of bank time deposit customers. x i express The i-th feature attribute in the model, where i is a positive integer and its value ranges from 1 to 16.
[0044] S2. Load the characteristic information of bank time deposit customers into a quantum state, specifically including S20 to S21:
[0045] S20. Determine the number of qubits to be used based on the characteristics of bank time deposit customers;
[0046] S21. Based on the number of qubits used, and using angle encoding, the characteristic information of bank time deposit customers is loaded into a quantum state.
[0047] 16-dimensional characteristic information of bank time deposit customers Each feature attribute in the matrix is calculated by taking its arctangent, i.e., by using the formula θ. i =arctan(x i ), to obtain the rotation angle θ of the quantum gate. i This represents the rotation angle of the quantum gate corresponding to each feature attribute. After obtaining the rotation angle of the quantum gate, it is used through R... y (θ) and R x (θ) A rotating quantum gate acts on a qubit. Here, 8 qubits are used to load 16-dimensional characteristic information of a bank's fixed deposit customer onto the amplitude of the quantum state. The specific quantum circuit is as follows: Figure 2 As shown.
[0048] S3. Use a quantum neural network to process the quantum state and obtain the prediction results of the characteristic information of bank time deposit customers. Construct a loss function based on the prediction results of the characteristic information of bank time deposit customers and the real labels of bank time deposit customers. Update the parameters of the quantum neural network according to the loss function until the termination condition is met.
[0049] Among them, the quantum neural network is used to process quantum states to obtain prediction results of characteristic information of bank time deposit customers, including:
[0050] S30. The quantum state is processed using a quantum neural network, the nth qubit of the quantum neural network is measured, and the prediction result of the characteristic information of the bank's time deposit customer is obtained based on the measurement result of the nth qubit.
[0051] The parameters of the quantum neural network are updated according to the loss function, including:
[0052] S31. Update the parameters of the quantum neural network according to the loss function, and use the parameter shifting rule to update the parameters of the quantum neural network.
[0053] Among them, quantum neural networks are constructed based on basic quantum operations supported by quantum computers. These basic quantum operations include: parameterized quantum gates R... x (θ), R y (θ), R z (θ) and the entangled quantum gate CZ, with the corresponding matrices as follows: The constructed quantum neural network, such as Figure 3 As shown.
[0054] Figure 3 In this context, RX and RY represent: an R is applied to each qubit. x (θ)R y (θ) Rotation gate, E represents a CZ quantum gate acting between two adjacent qubits, U is determined by R y (θ)R z (θ) Quantum gates constitute the network structure of U, as shown in... Figure 4 As shown.
[0055] Finally, a Pauli-Z measurement is performed on the fifth qubit. Based on the measurement results, the prediction results of the input bank time deposit customer's characteristic information are determined. The prediction results corresponding to measurement results less than 0 are normalized to 0, and the prediction results corresponding to measurement results greater than 0 are normalized to 1.
[0056] The characteristic information of bank time deposit customers in each batch b The data is input into the quantum neural network, and each bank's fixed deposit customer's feature information yields a corresponding prediction result y. Finally, all prediction results y in the batch are combined with the corresponding true label l indicating whether or not the customer agrees to the fixed deposit, to calculate the loss function characterizing the performance of the quantum neural network model. The expression for the mean squared error loss function is: Where y i This represents the prediction result of the quantum neural network for the characteristic information of the i-th bank's time deposit customer, l i This represents the real label corresponding to the feature information of the time deposit customer of bank i, and k is the number of feature information of the time deposit customer of bank i included in batch b.
[0057] The process of updating the parameters of the quantum neural network based on the loss function and then using the parameter shift rule is as follows:
[0058] A measurement operator In the parameterized quantum circuit U(θ) i The expected value function f(θ) under ) i ) can be represented as Then the expected value function f(θ) i Regarding parameterized quantum circuit parameters θ i gradient It can be represented as Expected value function f(θ) i U(θ) in ) i ) represents the parameterized quantum circuit corresponding to the quantum neural network, θ i This represents the parameters in a quantum neural network, and the method is called the parameter shift rule for calculating the gradient of the parameterized quantum circuit with respect to the operator expectation value.
[0059] The analytical gradient of the mean squared error loss function with respect to the parameters in the quantum neural network can be calculated using the parameter shift rule. Then, the parameters of the quantum neural network are updated using gradient descent on a classical computer. Finally, the network is trained for multiple epochs using a training set containing feature information from multiple bank time deposit customers until convergence. The performance of the trained quantum neural network, i.e., the latest version, is then tested using a test set.
[0060] S4. Classify the characteristic information of the preset users' bank time deposit customers based on the latest quantum neural network.
[0061] This invention constructs a highly efficient quantum neural network to segment target customer data for bank time deposits, thereby improving the success rate of bank time deposit marketing activities and significantly reducing marketing costs. Specifically:
[0062] This patent employs a self-designed quantum neural network. First, it downloads a dataset of bank time deposits and preprocesses it. Based on the dimensionality of the bank time deposit customer feature information, it selects the number of qubits to use. Then, using angle encoding, it loads the time deposit customer information into substates. Next, based on the basic quantum operations supported by quantum computers, it constructs a quantum neural network to process the quantum states loaded with the time deposit customer feature information. During this process, the dimensionality of the qubits is continuously reduced. Finally, it measures a single qubit and combines the measured information with the corresponding real labels of the time deposit customers to form a loss function. The parameters are continuously updated based on the loss function until a satisfactory threshold is reached. The beneficial effects are as follows:
[0063] This invention employs a novel computing paradigm based on the fundamental principles of quantum mechanics: quantum computing. The entanglement and superposition characteristics of quantum computing physical systems enable the learning of patterns in data that classical computers cannot learn. Furthermore, the powerful parallelism and non-locality of quantum neural networks allow the quantum neural network in this invention to outperform classical neural networks in extracting global features of bank deposit customer information and in terms of convergence speed, requiring less data for training. The design of the neural network model and the adjustment of hyperparameters are also simpler than with classical neural networks. This patent uses angle encoding to encode bank deposit customer information into quantum states, and the overall quantum neural network requires fewer layers and quantum operations. Therefore, the quantum neural network proposed in this patent can be applied to practical fields such as the classification of target customers for bank time deposits, using currently developed NISQ (noisy intermediate-scale quantum) computers. Currently, classical machine learning is mainly used to process bank time deposit data, while the task of classifying target customers for bank time deposit data using quantum computing has not yet been addressed.
[0064] In the above embodiments, although the steps are numbered S1, S2, etc., they are only specific embodiments given by the present invention. Those skilled in the art can adjust the execution order of S1, S2, etc. according to the actual situation, which is also within the protection scope of the present invention. It can be understood that in some embodiments, some or all of the above embodiments may be included.
[0065] An embodiment of the present invention provides a target customer classification system for bank time deposits based on quantum neural networks, comprising a data preprocessing module, a quantum state loading module, an update module, and a feature information classification module;
[0066] The data preprocessing module is used to preprocess the bank's time deposit dataset to obtain the characteristic information of bank time deposit customers;
[0067] The quantum state loading module is used to: load the characteristic information of bank time deposit customers into quantum states;
[0068] The update module is used to: process quantum states using a quantum neural network to obtain prediction results of the feature information of bank time deposit customers; construct a loss function based on the prediction results of the feature information of bank time deposit customers and the real labels of bank time deposit customers; and update the parameters of the quantum neural network based on the loss function until the termination condition is met.
[0069] The feature information classification module is used to classify the feature information of the preset users' bank time deposit customers based on the latest quantum neural network.
[0070] Optionally, in the above technical solution, the quantum state loading module is specifically used for:
[0071] The number of qubits to be used is determined based on the characteristics of bank time deposit customers.
[0072] Based on the number of qubits used and using angle encoding, the characteristic information of bank time deposit customers is loaded into quantum states.
[0073] Optionally, in the above technical solution, the updating module is further specifically used to: process the quantum state using a quantum neural network, measure the nth qubit of the quantum neural network, and obtain the prediction result of the characteristic information of the bank's time deposit customer based on the measurement result of the nth qubit.
[0074] Optionally, in the above technical solution, the updating module is further specifically used to: update the parameters of the quantum neural network according to the loss function, and update the parameters of the quantum neural network using the parameter shifting rule.
[0075] It should be noted that the beneficial effects of the bank time deposit target customer classification system based on quantum neural networks provided in the above embodiments are the same as those of the bank time deposit target customer classification method based on quantum neural networks, and will not be repeated here. Furthermore, the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, and will not be repeated here.
[0076] In another embodiment, the module includes a bank customer data preprocessing and quantum state encoding module, a quantum neural network construction module, a loss function construction module, and a quantum circuit parameter updating module. Specifically:
[0077] 1) The bank customer data preprocessing and quantum state encoding module is used to: represent the non-numeric categories of some features in the downloaded bank time deposit dataset with different numbers, and divide the dataset into training and testing datasets in a 9:1 ratio. Then, it is processed using R... y (θ) and R x (θ) A rotating quantum gate acts on a qubit (8 qubits are used here) to load 16-dimensional time deposit customer characteristic information onto the amplitude of the quantum state.
[0078] 2) The quantum neural network module is used to: construct a quantum neural network based on the characteristics of fixed deposit customer data, parameterized quantum gates supported by quantum computers, and entangled quantum gate operations to process the quantum states loaded with fixed deposit customer characteristic information in the previous module.
[0079] 3) The loss function module is used to: extract the characteristic data of time deposit customers in each batch b. The input is the quantum neural network built on the previous module. Each fixed deposit customer can obtain a predicted value y through the previous module. Then, by calculating the mean square error of the predicted values of all fixed deposit customers in each batch relative to their true labels, the loss function used to characterize the model performance is obtained.
[0080] 4) The quantum circuit parameter update module is used to: calculate the analytical gradient of the loss function of the previous module with respect to the quantum circuit parameters based on the existing parameter shift rule, then update the quantum circuit parameters using a classical computer, and finally train the fixed deposit customer training dataset for multiple epochs until the proposed method for classifying bank fixed deposit target customers based on quantum neural networks achieves the expected accuracy.
[0081] The following will provide a more detailed explanation.
[0082] Bank customer data preprocessing and quantum state encoding module:
[0083] First, download the Banking dataset. Then, represent the non-numeric categories of some features in the downloaded bank time deposit dataset using different numeric levels. Next, split the dataset into training and testing datasets in a 9:1 ratio. Then, process the 16-dimensional customer data from the dataset. The rotation angle of the quantum gate is obtained by taking the arctangent element by element, specifically through the formula θ. i =arctan(x i Obtain the rotation angle of the quantum gate, note the x here. i Let represent the i-th characteristic attribute of the client, where i indicates that the i-th rotating quantum gate accepts the i-th characteristic attribute of the client. Then, through R... y (θ) and R x (θ) A rotating quantum gate acts on a qubit (8 qubits are used here) to load 16-dimensional information about a time deposit customer onto the amplitude of the quantum state. The specific quantum circuit is as follows: Figure 2 As shown.
[0084] Constructing a quantum neural network module:
[0085] Based on parameterized quantum gates R supported by quantum computersx (θ)R y (θ)R z (θ) and the entangled quantum gate CZ, with the corresponding matrices as follows:
[0086]
[0087] A quantum neural network is constructed based on the quantum gate operations supported by the quantum computer to process the quantum states loaded with the characteristic information of fixed deposit customers in the previous module. The entire neural network structure is as follows: Figure 3 As shown. Figure 3 The encoder is a quantum state encoding module for the characteristic information of fixed deposit customers. RX and RY represent the operation of an R on each qubit. x (θ)R y (θ) Rotation gate, E represents the entanglement module indicating that a CZ quantum gate is applied between two adjacent qubits, and U is determined by R. y (θ)R z (θ) Quantum gates constitute its specific structure, as shown in... Figure 4 As shown.
[0088] Finally, a Pauli-Z measurement is performed on the fifth qubit to predict the input time deposit customer characteristic information. Results with measurements less than 0 are normalized to 0, and results with measurements greater than 0 are normalized to 1. A loss function module is constructed: the time deposit customer characteristic data from each batch b... The data is input into the quantum neural network built in the previous module (the quantum neural network construction module), and each fixed deposit customer receives a corresponding predicted value y. Finally, the predicted values y for all fixed deposit customers in the batch are combined with the corresponding true label l indicating whether or not they agree to the fixed deposit. The mean squared error loss function between the predicted value y and the corresponding label l is calculated, summed, and averaged to obtain the loss function used to characterize the performance of the quantum neural network model. The expression for the mean squared error loss function is:
[0089]
[0090] Among them, y i This represents the prediction made by the quantum neural network for current time deposit customers, l i This represents the true label of the current fixed deposit customer category, where k is the number of fixed deposit customers included in batch b.
[0091] Update quantum circuit parameter module:
[0092] First, a measurement operator In the parameterized quantum circuit U(θ) i The expected value under () can be expressed as: Then the expected value function f(θ) i Regarding the parameter θ of parameterized quantum circuits i The gradient can be expressed as: Expected value function f(θ) i U(θ) in ) i ) represents the parameterized quantum circuit corresponding to the quantum neural network, θ i The parameters represent the parameters of the parameterized quantum circuit, which are the rotation angles of the quantum gates in S21. The above method is called the parameter shift rule for calculating the gradient of the parameterized quantum circuit with respect to the operator expectation.
[0093] The mean squared error loss function is calculated with respect to the parameters in the quantum neural network using the parameter shift rule. Then, the parameters are updated using gradient descent with a classical computer. Finally, the model is trained for multiple epochs using a training dataset of time deposit customer characteristics until the proposed quantum neural network-based target customer classification method for bank time deposits converges. The trained model is then tested on a test set to evaluate its performance.
[0094] like Figure 5 As shown, an embodiment of the present invention provides a computer device 300, which includes a processor 320 coupled to a memory 310. The memory 310 stores at least one computer program 330, which is loaded and executed by the processor 320 to enable the computer device 300 to implement any of the above-mentioned methods for classifying target customers of bank time deposits based on quantum neural networks. Specifically:
[0095] The computer device 300 can vary considerably due to differences in configuration or performance. It may include one or more processors 320 (Central Processing Units, CPUs) and one or more memories 310. The one or more memories 310 store at least one computer program 330, which is loaded and executed by the one or more processors 320 to enable the computer device 300 to implement any of the quantum neural network-based target customer classification methods for bank time deposits provided in the above embodiments. Of course, the computer device 300 may also have wired or wireless network interfaces, a keyboard, and input / output interfaces for input and output. The computer device 300 may also include other components for implementing device functions, which will not be elaborated upon here.
[0096] An embodiment of the present invention provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to enable a computer to implement any of the above-mentioned methods for classifying target customers of bank time deposits based on quantum neural networks.
[0097] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.
[0098] In an exemplary embodiment, a computer program product or computer program is also provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform any of the aforementioned quantum neural network-based methods for classifying target customers for bank time deposits.
[0099] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.
[0100] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this disclosure can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product in one or more computer-readable media containing computer-readable program code.
[0101] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.
[0102] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.
Claims
1. A method for classifying target customers for bank time deposits based on quantum neural networks, characterized in that, include: Preprocess the bank time deposit dataset to obtain the feature information of bank time deposit customers; The characteristic information of the bank's time deposit customers is loaded into a quantum state; A quantum neural network is used to process quantum states to obtain prediction results of the characteristic information of bank time deposit customers. A loss function is constructed based on the prediction results of the characteristic information of bank time deposit customers and the real labels of bank time deposit customers. The parameters of the quantum neural network are updated according to the loss function until the termination condition is met. The system classifies the characteristic information of the preset users' bank time deposit customers based on the latest quantum neural network. Loading the feature information of the bank's time deposit customers into a quantum state includes: determining the number of qubits to be used based on the dimension of the bank's time deposit customers' feature information; and using angular encoding based on the number of qubits used. The processed 16-dimensional customer data The rotation angle of the quantum gate is obtained by taking the arctangent element by element, specifically through the formula... Obtain the rotation angle of the quantum gate. Let i represent the i-th characteristic attribute of the client, where i indicates that the i-th rotating quantum gate accepts the i-th characteristic attribute of the client, and then... and Rotating quantum gates act on qubits; here, 8 qubits are used to load 16-dimensional time deposit customer characteristic information onto the amplitude of the quantum state. Updating the parameters of the quantum neural network according to the loss function includes: updating the parameters of the quantum neural network according to the loss function, and updating the parameters of the quantum neural network using the parameter shifting rule; By processing quantum states using quantum neural networks, prediction results of characteristic information of bank time deposit customers are obtained, including: The quantum state is processed using a quantum neural network, the nth qubit of the quantum neural network is measured, and the prediction result of the characteristic information of the bank's time deposit customer is obtained based on the measurement result of the nth qubit. Among them, quantum neural networks are constructed based on basic quantum operations supported by quantum computers. These basic quantum operations include parameterized quantum gates. , , And the entangled quantum gate CZ, with the corresponding matrices as follows: , , , The constructed quantum neural network includes encoder, RX, first E, first RY, second E, second RY, third E, first U, second U, and third U; RX and RY represent: one function is applied to each qubit. Rotation gate, E represents a CZ quantum gate acting between two adjacent qubits, U includes the first one connected in sequence. , And the second The encoder is a quantum state encoding module used to load the feature information of the bank's fixed deposit customers into a quantum state. The encoder's input data consists of 8 qubits, and its output data is the input data of RX. The output data of RX is the input data of the first E. The output data of the first E consists of the data corresponding to the qubits other than the first and last qubits, which serves as the input data of the first RY. The output data of the first RY is the input data of the second E. The output data of the second E consists of the data corresponding to the qubits other than the second and penultimate qubits, which serves as the input data of the second RY. The output data of the second RY serves as the input data of the third E. The data corresponding to the fourth qubit in the output data of the third E serves as the input data of the first U. The data corresponding to the fifth qubit in the output data of the third E serves as the input data of the second U. Then, the output data of the first U and the output data of the second U are transformed through a CZ quantum gate. Finally, the output data of the second U is used as the input data of the third U. The output data of the third U is the output data to be measured of the quantum neural network.
2. A target customer classification system for bank time deposits based on quantum neural networks, characterized in that, It includes a data preprocessing module, a quantum state loading module, an update module, and a feature information classification module; The data preprocessing module is used to: preprocess the bank's time deposit dataset to obtain the feature information of bank time deposit customers; The quantum state loading module is used to: load the feature information of the bank's time deposit customers into quantum states; The update module is used to: process the quantum state using a quantum neural network to obtain the prediction result of the feature information of bank time deposit customers; construct a loss function based on the prediction result of the feature information of bank time deposit customers and the real labels of bank time deposit customers; and update the parameters of the quantum neural network based on the loss function until the termination condition is met. The feature information classification module is used to classify the feature information of the preset user's bank time deposit customers according to the latest obtained quantum neural network; The quantum state loading module is specifically used for: The number of qubits to be used is determined based on the characteristics of bank time deposit customers. Based on the number of qubits used and using angle encoding, the characteristic information of the bank's time deposit customers is loaded into a quantum state; The processed 16-dimensional customer data The rotation angle of the quantum gate is obtained by taking the arctangent element by element, specifically through the formula... Obtain the rotation angle of the quantum gate. Let i represent the i-th characteristic attribute of the client, where i indicates that the i-th rotating quantum gate accepts the i-th characteristic attribute of the client, and then... and Rotating quantum gates act on qubits; here, 8 qubits are used to load 16-dimensional time deposit customer characteristic information onto the amplitude of the quantum state. The update module is also specifically used to: process the quantum state using a quantum neural network, measure the nth qubit of the quantum neural network, and obtain the prediction result of the characteristic information of the bank's time deposit customer based on the measurement result of the nth qubit; The update module is also specifically used to: update the parameters of the quantum neural network according to the loss function, and update the parameters of the quantum neural network using the parameter shifting rule; Among them, quantum neural networks are constructed based on basic quantum operations supported by quantum computers. These basic quantum operations include parameterized quantum gates. , , And the entangled quantum gate CZ, with the corresponding matrices as follows: , , , The constructed quantum neural network includes encoder, RX, first E, first RY, second E, second RY, third E, first U, second U, and third U; RX and RY represent: one function is applied to each qubit. Rotation gate, E represents a CZ quantum gate acting between two adjacent qubits, U includes the first one connected in sequence. , And the second The encoder is a quantum state encoding module used to load the feature information of the bank's fixed deposit customers into a quantum state. The encoder's input data consists of 8 qubits, and its output data is the input data of RX. The output data of RX is the input data of the first E. The output data of the first E consists of the data corresponding to the qubits other than the first and last qubits, which serves as the input data of the first RY. The output data of the first RY is the input data of the second E. The output data of the second E consists of the data corresponding to the qubits other than the second and penultimate qubits, which serves as the input data of the second RY. The output data of the second RY serves as the input data of the third E. The data corresponding to the fourth qubit in the output data of the third E serves as the input data of the first U. The data corresponding to the fifth qubit in the output data of the third E serves as the input data of the second U. Then, the output data of the first U and the output data of the second U are transformed through a CZ quantum gate. Finally, the output data of the second U is used as the input data of the third U. The output data of the third U is the output data to be measured of the quantum neural network.
3. A computer device, characterized in that, The computer device includes a processor coupled to a memory storing at least one computer program, which is loaded and executed by the processor to enable the computer device to implement the method for classifying target customers for bank time deposits based on a quantum neural network as described in claim 1.
4. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one computer program, which is loaded and executed by a processor to enable the computer to implement the method for classifying target customers for bank time deposits based on a quantum neural network as described in claim 1.