A quantum evidence reasoning network pattern recognition method, system, device and storage medium based on quantum evidence theory

Abstract

The application relates to the technical field of quantum computing, and relates to a quantum evidence reasoning network pattern recognition method, system and device based on quantum evidence theory and a storage medium; the quantum evidence reasoning network pattern recognition method based on quantum evidence theory comprises the following steps: acquiring a classical feature data set; converting classical data into an initial quantum state and distributing the initial quantum state into a plurality of independent category channels to extract multi-qubit features; generating a quantum basic probability amplitude function with a complex phase based on the multi-qubit features; performing quantum tensor product operation on the quantum basic probability amplitude functions generated by each quantum channel to fuse and form a composite quantum state representing a global state; measuring the appearance frequency of a specific ground state in the composite quantum state, calculating the quantum basic probability distribution of each data category based on a quantum evidence combination rule, and selecting a data category with the maximum quantum basic probability as a recognition result. The application can improve the accuracy of pattern classification.

Description

Technical Field

[0001] This invention relates to the field of quantum computing technology, and in particular to a method, system, device and storage medium for quantum evidence reasoning network pattern recognition based on quantum evidence theory. Background Technology

[0002] Pattern recognition plays a crucial role in data processing tasks. To address the uncertainty in data, some quantum classification models have been applied to relevant scenarios. However, in practical computing environments, these models still have several significant limitations in their structural design. First, due to the lack of an end-to-end execution mechanism, models often rely on externally provided known samples when calculating data trust levels, typically using the cosine similarity between test samples and evidence source centers to generate probability assignments. This approach is highly inflexible when dealing with complex data distributions. Second, existing computational processes suffer from cumbersome data interaction. In a non-end-to-end framework, intermediate data needs to be frequently transferred between classical computing devices and quantum hardware. This cross-domain data reading and interaction process not only increases system latency but also introduces significant computational errors. The interruption of computational continuity weakens the parallel processing advantages inherent in quantum computing, resulting in generally low operating efficiency of existing models in complex environments. Third, existing processing methods have significant shortcomings in terms of information representation dimensions. Traditional evidence theory typically defines basic probabilistic assignment functions only in the real number domain, relying solely on the magnitude of real numbers to characterize trust levels, while completely ignoring the complex phase information that should be included in quantum states. In quantum computing systems, phase can record the correlation characteristics between different states; the lack of phase causes a significant loss of data information during the combination of different evidence sources, making it difficult for existing classification models to provide rich mathematical expressions and effectively suppress conflicting information in multi-source data. Summary of the Invention

[0003] This application aims to at least solve the technical problems existing in the prior art, and to provide a method, system, device and storage medium for quantum evidence reasoning network pattern recognition based on quantum evidence theory.

[0004] In a first aspect, the present invention provides a quantum evidence reasoning network pattern recognition method based on quantum evidence theory, the method comprising: Obtain a classic feature dataset, which includes standardized classic features from multiple different data categories; The number of qubits in each category channel is determined based on the number of available qubits and the total number of data categories; an independent category channel is assigned to each data category, and the normalized classical features are mapped to a rotation angle through an inverse trigonometric function. The rotation angle is passed as a parameter to the rotation gate, and the classical data is encoded into an initial quantum state using the rotation gate; the number of available qubits is the maximum number of physical qubits that the quantum computing hardware system executing this method can actually participate in quantum state encoding and circuit operations. The initial quantum state is input into a variable quantum circuit, which contains a multi-layer structure. In each layer, parameterized single-bit rotation gates and controlled NOT gates are used alternately to extract the correlation features in each channel to obtain multi-qubit features. Then, the multi-qubit features are compressed to the last single bit through quantum pooling operation. An output rotation gate is applied to the last single bit to generate a quantum fundamental probability amplitude function carrying a complex phase. The quantum tensor product operation is performed on the quantum fundamental probability amplitude functions generated by each quantum channel, and the functions are fused to form a composite quantum state that represents the global state; The frequency of occurrence of a specific ground state in the composite quantum state is measured, and the quantum fundamental probability distribution of each data category is calculated based on the quantum evidence combination rule. The data category with the highest quantum fundamental probability is selected as the identification result.

[0005] Secondly, the present invention provides a quantum evidence reasoning network pattern recognition system based on quantum evidence theory, the system comprising: The acquisition module is used to acquire a classic feature dataset, which includes standardized classic features of multiple different data categories; The processing module is used to determine the number of qubits in each category channel based on the number of available qubits and the total number of data categories; allocate an independent category channel for each data category; map the normalized classical features to a rotation angle through an inverse trigonometric function; pass the rotation angle as a parameter to the rotation gate; and use the rotation gate to encode the classical data into an initial quantum state; the number of available qubits is the maximum number of physical qubits that the quantum computing hardware system executing this method can actually participate in quantum state encoding and circuit operations. The feature extraction module is used to input the initial quantum state into the variable quantum circuit. The variable quantum circuit contains a multi-layer structure. In each layer, parameterized single-bit rotation gates and controlled NOT gates are used alternately to extract the correlation features in each channel to obtain multi-qubit features. Then, the multi-qubit features are compressed to the last single bit through quantum pooling operation. An output rotation gate is applied to the last single bit to generate a quantum fundamental probability amplitude function carrying a complex phase. The fusion module is used to perform quantum tensor product operations on the quantum fundamental probability amplitude functions generated by each quantum channel and fuse them to form a composite quantum state that represents the global state. The output module is used to measure the frequency of occurrence of a specific ground state in the composite quantum state, calculate the quantum fundamental probability distribution of each data category based on the quantum evidence combination rule, and select the data category with the highest quantum fundamental probability as the identification result.

[0006] Thirdly, the present invention provides an electronic device, the electronic device comprising: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the quantum evidence reasoning network pattern recognition method based on quantum evidence theory described above.

[0007] Fourthly, the present invention also provides a computer-readable storage medium storing at least one computer program, which is executed by a processor in an electronic device to implement the above-described quantum evidence reasoning network pattern recognition method based on quantum evidence theory.

[0008] In summary, this application includes the following beneficial technical effects: This invention constructs an end-to-end learning architecture that integrates classical data encoding, variable quantum feature extraction, multi-channel evidence fusion, and classification decision-making within the same quantum circuit framework. It allocates independent channels to each data category and adaptively calculates the number of qubits in each channel based on the available qubits to optimize quantum resource allocation. Simultaneously, it directly generates quantum fundamental probability amplitude functions carrying complex phases at the end of each channel. This ensures that the modulus and phase information in the complex amplitude values ​​are fully preserved and participate in interference operations during subsequent multi-channel quantum tensor product fusion. Furthermore, it calculates a normalized quantum fundamental probability distribution based on quantum evidence combination rules to achieve fusion decision-making. Compared to existing methods that define probability distributions only in the real domain and require frequent classical-quantum data switching, this invention effectively reduces system errors and operational delays introduced by cross-domain data conversion, enriches the information dimension of trust expression, enhances the coherence of consistent evidence, and cancels out conflicting evidence, significantly improving the model's classification robustness under evidence conflict and noise interference. Attached Figure Description

[0009] Figure 1 A flowchart illustrating a quantum evidence reasoning network pattern recognition method based on quantum evidence theory, provided in an embodiment of the present invention; Figure 2 A system structure diagram of a quantum evidence reasoning network pattern recognition method based on quantum evidence theory provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the quantum state angle encoding circuit structure in this application; Figure 4 This is a schematic diagram of the single-layer feature extraction variable quantum circuit structure in this application; Figure 5 This is a schematic diagram of the positively controlled NOT gate connection structure in this application; Figure 6 This is a schematic diagram of the reverse controlled NOT gate connection structure in this application; Figure 7 This is a schematic diagram of the quantum pooling structure in this application; Figure 8 This is a schematic diagram of the generation structure of the quantum fundamental probability amplitude function in this application; Figure 9 This is a schematic diagram of the quantum circuit structure for evidence combination in this application; Figure 10 This is a schematic diagram showing an experimental comparison between the quantum evidence reasoning network algorithm in this application and the MORE algorithm, VSQL algorithm, and ordinary quantum neural network algorithm; Figure 11 This is a schematic diagram of the ablation experiment of the quantum evidence reasoning network algorithm in this application; Figure 12 This is a schematic diagram of the structure of an electronic device that implements the quantum evidence reasoning network pattern recognition method based on quantum evidence theory, according to an embodiment of the present invention.

[0010] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0011] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0012] In the description of this invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0013] In the description of this invention, unless otherwise specified and limited, it should be noted that the terms "installation", "connection" and "linking" should be interpreted broadly. For example, they can refer to mechanical or electrical connections, or internal connections between two components. They can be direct connections or indirect connections through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0014] Reference Figure 1 and Figure 2 In this embodiment, the quantum evidence reasoning network pattern recognition method based on quantum evidence theory includes: S1. Obtain the classic feature dataset.

[0015] The classic feature dataset includes standardized classic features from multiple different data categories.

[0016] Specifically, the steps for obtaining a classic feature dataset are as follows: S11. Obtain multiple initial classic features, wherein the multiple initial classic features belong to different data categories, and each method data category includes at least one initial classic feature; S12. Principal component analysis is used to reduce the dimensionality and standardize the classic feature dataset to obtain standardized classic features. S13. Integrate the standardized classic features to obtain the final classic feature dataset.

[0017] In practice, the collected classical feature data is first subjected to dimensionality reduction operation through principal component analysis algorithm to adapt to limited quantum hardware resources.

[0018] S2. Determine the number of qubits for each category channel based on the number of available qubits and the total number of data categories; assign an independent category channel to each data category, map the normalized classical features to a rotation angle using an inverse trigonometric function, pass the rotation angle as a parameter to the rotation gate, and use the rotation gate to encode the classical data into an initial quantum state.

[0019] The number of available qubits is the maximum number of physical qubits that the quantum computing hardware system executing this method can actually participate in quantum state encoding and circuit operations.

[0020] To avoid disruptions to circuit operation due to high-dimensional features, a dimensionality reduction operation is performed. The number of qubits that can be allocated to each category of channel is pre-determined based on the maximum number of qubits supported by the hardware and the total number of channels of each category. The formula for calculating the number of qubits per category of channel is: ; in, The number of qubits for each type of channel, For the number of data categories, Indicates rounding down; This indicates taking the maximum value.

[0021] Reading feature values Then, the standardized classical features are mapped to rotation angles using inverse trigonometric functions as follows: ; in, Represents the first in the classic feature dataset A standardized classic feature The A specific feature value, Standardize the classic feature index in the classic feature dataset. For standardization of classic features Index of feature values ​​in the middle. Represents the arcsine function. Representing characteristic values The rotation angle value obtained after mapping with inverse trigonometric functions; Subsequently, , and Operations are applied sequentially, and feature data is input into the corresponding channels. Quantum circuits, such as... Figure 3 Classical samples are transformed into the initial quantum system state. Its mathematical derivation formula is: ; Representing the Classic sample data The initial quantum state is established after a complete angle encoding operation; The quantum tensor product operator is used to combine multiple independent qubit states into a larger global composite quantum state. This indicates the number of data categories, i.e., the number of channels; The index represents the category channel; the network assigns a unique quantum channel to each category. Representative at the Index of qubits within each category channel , , The quantum fundamental rotation gates, representing rotations around the X, Y, and Z axes of the Bloch sphere, respectively, have the following characteristic values. and its mapping angle These parameters are input into these rotating gates, thereby deflecting the quantum state at a specific angle; Representing the In the first category channel, the... The initial basis state of 1 qubit.

[0022] S3. Input the initial quantum state into the variable quantum circuit, which contains a multi-layer structure. In each layer, parameterized single-bit rotation gates and controlled NOT gates are used alternately to extract the correlation features in each channel to obtain multi-qubit features. Then, the multi-qubit features are compressed to the last single bit through quantum pooling operation. An output rotation gate is applied to the last single bit to generate a quantum fundamental probability amplitude function carrying a complex phase.

[0023] In this embodiment, the initial quantum state is represented as a bit sequence consisting of multiple qubits, wherein the number of qubits in the bit sequence is... The quantum basic probability amplitude function (QBPA) in quantum evidence theory characterizes the uncertainty of data and expresses the degree of support for the discriminative space, similar to classical Bayesian probability. After the initial quantum state is constructed, the data enters a variable quantum circuit containing L layers of trainable logic gates. The system structure of the variable quantum circuit is as follows: Figure 4 As shown.

[0024] In each layer, parameterized single-qubit rotation gates and controlled NOT gates are used alternately to extract correlation features within each channel, resulting in multi-qubit features, including: S31. Apply a parameterized single-bit rotation gate to each qubit corresponding to the initial quantum state; S32. Apply controlled NOT gates sequentially along the forward direction of the bit sequence, making the first bit... The number of qubits controls the first One quantum bit, ; The controlled-NOT gate, also known as the CNOT gate, is a fundamental two-qubit logic gate in quantum computing. Information from the preceding bit is passed forward through the controlled-NOT gate, establishing a forward connection. Quantum circuits with forward CNOT gates, such as Figure 5 This facilitates data mixing, and the corresponding operation can be represented as:

[0025] in, Represents a forward connection operation block in a variable quantum circuit, indicating a series of entangled logic combinations executed sequentially; The chain multiplication symbol, representing quantum operators, in quantum circuits indicates that the related logic gate operations are performed according to indexes. The increasing order is applied sequentially; This represents the total number of qubits currently allocated in the independent channels; The index number representing the qubit being operated on; and Representing the first The first qubit and the next one One qubit. In this expression, the one that comes first... Defined as control bits, and listed later. Defined as the target bit. The logic judgment rules of the gate circuit corresponding to the controlled NOT gate are relatively straightforward. When the control bit... In the ground state At that time, target bit The state will remain unchanged; when the control bits In the ground state At that time, target bit The state will be flipped (i.e.) Flipped to , Flipped to When a qubit is in a superposition state, this conditional flip creates quantum entanglement between the two bits. In feature extraction circuits, controlled NOT gates are primarily used to establish physical connections between different feature dimensions (i.e., different qubits). By allowing the index to range from 1 to... This operation is performed sequentially, and the feature information in the preceding bits is smoothly passed to the following bits, where it mixes with the information in the following bits. The potential correlation patterns between the data are extracted by the network in this way.

[0026] S33. Apply parameterized single-bit rotation gates to each qubit again; S34. Apply the controlled NOT gate sequentially in reverse order along the bit sequence, making the first bit... The number of qubits controls the first One quantum bit; The information of the last bit is passed forward through a controlled NOT gate, that is, a reverse connection is established. Quantum circuits with inverse CNOT gates, such as Figure 6 This further promotes quantum entanglement, and this part of the operation can be represented as:

[0027] Finally, the third application The gate is used to complete the feature extraction of this layer. Through this interlaced rotation and entanglement operation, the potential correlation patterns between data are effectively captured by the system.

[0028] S35. A parameterized single-qubit rotation gate is applied to each qubit for the third time to complete the correlation feature extraction of the current layer, and the multi-qubit features are determined based on the extracted correlation features.

[0029] The parameterized single-bit rotation gate in this embodiment is Door, A controlless NOT gate is a single-qubit rotation gate that contains three optimizable variable parameters to achieve parameterized mapping. In pattern recognition tasks, this is similar to the independent weight adjustments and nonlinear transformations applied to a single feature node in a traditional neural network. A controlled-NOT gate (also known as a CNOT gate) is a two-qubit logic gate. The operation of a CNOT gate depends on the cooperation of two qubits: one as the control bit and the other as the target bit. The state of the target bit is only flipped when the control bit satisfies a specific state. CNOT gates do not have adjustable parameters, but they can force a physical connection between two independent qubits, breaking the single-dimensional limitation and allowing information from different feature channels to interfere and mix—similar to establishing connections between multiple feature nodes in a traditional neural network. If the variational circuit only contains... Without gates, each feature dimension will always operate independently, and multi-source information cannot be fused; if relying solely on CNOT gates, the system will lose its ability to learn autonomously and adjust errors. The alternating use of parameterized single-bit rotation gates and controlled-NOT gates allows the single-dimensional features of the qubit to be parameterized first, then immediately correlated and mixed with features of other dimensions using controlled-NOT gates, followed by parameterized single-bit rotation gates to fine-tune the new mixed state. This alternating design constructs a trainable network with high expressive power. Furthermore, the controlled-NOT gate acts as a bridge for information transmission, passing the information extracted from the preceding bits to the following; subsequently, The gate performs weighted compression and refinement on the newly gathered information. Through multiple such alternating cycles, the high-dimensional features originally scattered across multiple qubits are passed layer by layer like a funnel and relatively smoothly condensed into a single qubit at the end. This is achieved using the CNOT gate and... Through the alternating action of the gates, the characteristic information within the channel is gradually gathered to the single quantum bit at the end. Above. Specific output. The gate is subjected to a quantum fundamental probability amplitude function containing a complex phase on the last bit.

[0030] In the independent channels of each category, CNOT gates and The gates are called alternately, and the system executes a total of The next iteration. The characteristic information within each channel is gradually compressed and converged to a single qubit at the end. Above. Quantum circuits for quantum pooling, such as... Figure 7 The multiplication formula for quantum pooling is defined as:

[0031] It is in the The application on each qubit The parameters of the gate, here The gate is a variable quantum gate, that is... These are learnable parameters, similar to the parameters between neurons in a classic neural network. After compression is complete, the system stores the last bit. The last time a specific output was applied Gate, Quantum Fundamental Probability Amplitude Function (QBPA) Consequently, the quantum circuitry of the quantum fundamental probability amplitude function is generated, such as... Figure 8 .

[0032] Specifically, the generation formula for the quantum fundamental probability amplitude function is as follows: ; in, Representing the zero state, on the Bloch sphere, Pointing in the positive direction of the Z-axis; Representing an excited state, on a Bloch sphere, Pointing in the negative direction of the Z-axis; Denotes the quantum fundamental probability amplitude function of the entire set. Indicates the first The basic probability amplitude function of a category; ground state The quantum fundamental probability amplitude function of the entire set is stored. ground state Stored the first Basic probability amplitude function of the category , Representative identification framework complete set.

[0033] Because the fundamental probability amplitude function is constructed in complex form, not only does the amplitude reflect the confidence level, but the complex phase is also preserved. This complex phase-based representation allows for a better reflection of the state differences in the confidence level distribution.

[0034] S4. Perform quantum tensor product operation on the quantum fundamental probability amplitude functions generated by each quantum channel and fuse them to form a composite quantum state that represents the global state.

[0035] After obtaining After the independent quantum fundamental probability amplitude functions within each channel are obtained, the quantum tensor product operation is used to combine the quantum states of each channel.

[0036] Global state, representing the overall state of the system It is formed, as shown in the following formula:

[0037] Indicates the first The category to which each independent quantum channel belongs. Representing the In the quantum fundamental probability amplitude function (QBPA) generated by each independent quantum channel, the corresponding support class The operation, by maintaining the confidence level of the evidence source itself, mathematically preserves the independence of the evidence source.

[0038] S5. Measure the frequency of occurrence of a specific ground state in the composite quantum state, calculate the quantum fundamental probability distribution of each data category based on the quantum evidence combination rule, and select the data category with the highest quantum fundamental probability as the identification result.

[0039] To calculate the category to which a sample belongs The probability is used to observe the fused system states. A specific ground state. (Only the first) The frequency of occurrence of (bit 1, all others 0) is statistically analyzed by the measurement module, and the corresponding quantum circuit in the evidence statistics section is as follows: Figure 9 Based on the Quantum Evidence Combination Rule (QECR), the numerator term of a single class is calculated by dividing it by the sum of the numerator terms of all classes, resulting in the final quantum fundamental probability distribution (QBPD). Specifically, the calculated expression for the quantum fundamental probability distribution is:

[0040] in, The representative sample belongs to the channel category. The final quantum fundamental probability distribution, This indicates that the sample is divided into the first... A single-category proposition for a class. This indicates that the sample is divided into the first... A single-category proposition specific to a class; The total number of target categories in a pattern classification task; , , The value range is from 1 to ,in This is the category index where the target probability is currently being calculated. It is the category index used in the denominator for iterating and summing. This represents the index of auxiliary channels other than the target category; Representing the In the quantum fundamental probability amplitude function (QBPA) generated by each independent quantum channel, the corresponding support class The inherent trust level amplitude; Representing the In the quantum fundamental probability amplitude function generated by each quantum channel, corresponding to the entire set, The trust level amplitude; This represents the chain multiplication operator.

[0041] In a preferred embodiment of this example, the prediction error is evaluated using the classification cross-entropy function based on historical recognition results. The gradient is calculated on the quantum circuit using the parameter-shift rule to obtain the current error solution gradient. Then, the parameters are updated on a classical computer using the Adam optimization algorithm. Specifically, all trainable parameters (i.e., each...) are calculated. The gradient value (the rotation angle of the gate) is fed back to the classical computer after the gradient of the current step. The Adam optimizer uses these gradients to adjust the parameters according to its classical update rules, thereby optimizing the various parameters in the quantum circuit. The rotation parameters of the door are updated in reverse accordingly.

[0042] To verify the effectiveness of this invention, nine publicly available datasets from different fields were selected for testing. These datasets, sourced from the publicly available UCI Machine Learning Repository, cover real-world scenarios such as medical diagnosis, physical signals, and plant classification, specifically including the Iris, Wine, Breast Cancer, Banknote, Ionosphere, Sonar, Parkinsons, Blood Transfusion, and Rice datasets. Ten replicate experiments were used as the evaluation mechanism. The method of this invention (i.e., the quantum evidence reasoning network) was compared with three existing end-to-end quantum algorithms: the MORE algorithm, the VSQL algorithm, and a conventional quantum neural network (QNN) algorithm.

[0043] Experimental results show that the proposed method significantly improves the overall classification accuracy on most datasets. In a comparative analysis of the nine datasets mentioned above, the algorithm of this invention achieved relatively superior classification results on seven datasets: Iris, Wine, Banknote, Sonar, Parkinsons, Blood Transfusion, and Rice. Specific statistical data, such as... Figure 10This invention achieves a classification accuracy of 100.00% on the Iris dataset, which has relatively well-defined features, and 80.00% on the Wine dataset, which has more feature dimensions. In comparison, the VSQL algorithm achieves only 55.64% accuracy on the Banknote dataset, and the MORE algorithm achieves only 64.10% accuracy on the Rice dataset. This objective comparison demonstrates that the end-to-end generalized quantum architecture proposed in this invention has relatively good advantages in complex feature extraction.

[0044] Furthermore, the effectiveness and rationality of the internal core components of this invention were further verified through ablation experiments. The experiments tested the difference in classification accuracy between the complete model of this invention and a variant model with different core modules removed. Specific test data were recorded, such as... Figure 11 When processing the Wine dataset, the complete algorithm of this invention achieved an accuracy of 80.00%. If the multi-channel structure was removed, the accuracy dropped to 74.44%; if the evidence generation module was removed, the accuracy dropped to 76.11%; and when both the multi-channel and evidence generation modules were removed, the accuracy of the base network was only 75.56%. On the Iris dataset, the accuracy of the double-missing architecture also dropped from 100.00% to 96.67%. The loss of a single module also weakens the overall discriminative ability of the system. These numerical comparisons intuitively demonstrate that the multi-channel parallel extraction mechanism and the independent evidence generation module play a crucial synergistic role in improving the overall classification performance of the model.

[0045] In summary, this invention proposes an end-to-end quantum evidence reasoning network classification method. It demonstrates good accuracy and stability in data feature extraction and the fusion of uncertain information.

[0046] Based on the same inventive concept, one embodiment of the present invention proposes a quantum evidence reasoning network pattern recognition system based on quantum evidence theory.

[0047] The quantum evidence reasoning network pattern recognition system based on quantum evidence theory described in this invention can be installed in an electronic device. Depending on the functions implemented, the quantum evidence reasoning network pattern recognition system based on quantum evidence theory includes: The acquisition module is used to acquire a classic feature dataset, which includes standardized classic features of multiple different data categories; The processing module is used to determine the number of qubits in each category channel based on the number of available qubits and the total number of data categories; allocate an independent category channel for each data category; map the normalized classical features to a rotation angle through an inverse trigonometric function; pass the rotation angle as a parameter to the rotation gate; and use the rotation gate to encode the classical data into an initial quantum state; the number of available qubits is the maximum number of physical qubits that the quantum computing hardware system executing this method can actually participate in quantum state encoding and circuit operations. The feature extraction module is used to input the initial quantum state into the variable quantum circuit. The variable quantum circuit contains a multi-layer structure. In each layer, parameterized single-bit rotation gates and controlled NOT gates are used alternately to extract the correlation features in each channel to obtain multi-qubit features. Then, the multi-qubit features are compressed to the last single bit through quantum pooling operation. An output rotation gate is applied to the last single bit to generate a quantum fundamental probability amplitude function carrying a complex phase. The fusion module is used to perform quantum tensor product operations on the quantum fundamental probability amplitude functions generated by each quantum channel and fuse them to form a composite quantum state that represents the global state. The output module is used to measure the frequency of occurrence of a specific ground state in the composite quantum state, calculate the quantum fundamental probability distribution of each data category based on the quantum evidence combination rule, and select the data category with the highest quantum fundamental probability as the identification result.

[0048] The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0049] The various variations and specific examples of the quantum evidence reasoning network pattern recognition method based on quantum evidence theory provided in the above embodiments are also applicable to the quantum evidence reasoning network pattern recognition system based on quantum evidence theory in this embodiment. Through the foregoing detailed description of the quantum evidence reasoning network pattern recognition method based on quantum evidence theory, those skilled in the art can clearly understand the implementation method of the quantum evidence reasoning network pattern recognition system based on quantum evidence theory in this embodiment. For the sake of brevity, it will not be described in detail here.

[0050] This application also discloses an electronic device, with reference to Figure 12 The electronic device may include at least one processor 10, a memory 11 communicatively connected to at least one processor, a communication bus 12, and a communication interface 13. It may also include a computer program stored in the memory 11 and executable on the processor 10, such as a method program for quantum evidence reasoning network pattern recognition based on quantum evidence theory.

[0051] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., executing a method for pattern recognition in quantum evidence reasoning networks based on quantum evidence theory), and calls data stored in the memory 11 to perform various functions of the electronic device and process data.

[0052] The memory 11 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of an electronic device, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device, such as a plug-in portable hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory 11 can include both internal and external storage units of the electronic device. The memory 11 can be used not only to store application software and various types of data installed on the electronic device, such as the code of a quantum evidence reasoning network pattern recognition method program based on quantum evidence theory, but also to temporarily store data that has been output or will be output.

[0053] The communication bus 12 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0054] Communication interface 13 is used for communication between the aforementioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, Bluetooth interface, etc.), typically used to establish communication connections between the electronic device and other electronic devices. The user interface may be a display, an input unit (such as a keyboard), and optionally, a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface.

[0055] The electronic device may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to at least one processor via a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power sources, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be elaborated further here.

[0056] It should be understood that the embodiments are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0057] Furthermore, if the modules / units integrated into the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile.

[0058] This application provides a computer-readable storage medium, including, for example, any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM). The computer-readable storage medium stores a computer program capable of being loaded by a processor and executing the quantum evidence reasoning network pattern recognition method based on quantum evidence theory described in the above embodiments.

[0059] In the description of this specification, the references to terms such as "an embodiment," "some embodiments," "example," "specific example," "a implementation," "a preferred implementation," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0060] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A quantum evidence reasoning network pattern recognition method based on quantum evidence theory, characterized in that, The method is applied to pattern recognition tasks in image recognition, medical diagnosis, or physical signal classification. The method includes: Obtain a classic feature dataset, which includes standardized classic features of multiple different data categories. The classic feature data is image data, medical detection data, or physical sensor data to be identified. The number of qubits in each category channel is determined based on the number of available qubits and the total number of data categories; an independent category channel is assigned to each data category, and the normalized classical features are mapped to a rotation angle through an inverse trigonometric function. The rotation angle is passed as a parameter to the rotation gate, and the classical data is encoded into an initial quantum state using the rotation gate; the number of available qubits is the maximum number of physical qubits that the quantum computing hardware system executing this method can actually participate in quantum state encoding and circuit operations. The initial quantum state is input into a variable quantum circuit, which contains a multi-layer structure. In each layer, parameterized single-bit rotation gates and controlled NOT gates are used alternately to extract the correlation features in each channel to obtain multi-qubit features. Then, the multi-qubit features are compressed to the last single bit through quantum pooling operation. An output rotation gate is applied to the last single bit to generate a quantum fundamental probability amplitude function carrying a complex phase. The quantum tensor product operation is performed on the quantum fundamental probability amplitude functions generated by each quantum channel, and the functions are fused to form a composite quantum state that represents the global state; Measure the frequency of occurrence of a specific ground state in the composite quantum state, and calculate the quantum fundamental probability distribution for each data category based on the following quantum evidence combination rule: ； in, The representative sample belongs to the channel category. The final quantum fundamental probability distribution, This indicates that the sample is divided into the first... A single category of a class. This indicates that the sample is divided into the first... A single category of a class; The total number of target categories in a pattern classification task; , , The value range is from 1 to ,in This is the category index where the target probability is currently being calculated. It is the category index used in the denominator for iterating and summing. This represents the index of auxiliary channels other than the target category; Representing the In the quantum fundamental probability amplitude function (QBPA) generated by each independent quantum channel, corresponding to the support class... The inherent trust level amplitude; Representing the In the quantum fundamental probability amplitude function generated by each quantum channel, corresponding to the entire set, The trust level amplitude; Represents the chain multiplication operator, where the specific ground state is only the first... The ground state in which one quantum bit is 1 and the rest of the quantum bits are 0; The data category with the highest quantum fundamental probability is selected as the identification result.

2. The quantum evidence reasoning network pattern recognition method based on quantum evidence theory as described in claim 1, characterized in that, The steps for obtaining the classic feature dataset are as follows: Multiple initial classic features are obtained, and the multiple initial classic features belong to different data categories. Each method data category includes at least one initial classic feature. Principal component analysis is used to reduce the dimensionality and standardize the classical feature dataset to obtain standardized classical features. The final classic feature dataset is obtained by integrating standardized classic features.

3. The quantum evidence reasoning network pattern recognition method based on quantum evidence theory as described in claim 1, characterized in that, The formula for calculating the number of qubits in each channel category is: ； in, The number of qubits for each type of channel, For the number of data categories, Indicates rounding down; This indicates taking the maximum value.

4. The quantum evidence reasoning network pattern recognition method based on quantum evidence theory as described in claim 3, characterized in that, The method of mapping standardized classical features to rotation angles using inverse trigonometric functions is as follows: ； in, Represents the first in the classic feature dataset A standardized classic feature The A specific feature value, Standardize the classic feature index in the classic feature dataset. For standardization of classic features Index of feature values ​​in the middle. Represents the arcsine function. Representing characteristic values The rotation angle value obtained after mapping with inverse trigonometric functions; The initial quantum state is constructed as follows: ； Representing the Classic sample data The initial quantum state is established after a complete angle encoding operation; This represents the quantum tensor product operator. Index representing the category channel, Representative at the Index of qubits within each channel category , , These represent quantum-based rotating gates that rotate around the X, Y, and Z axes of the Bloch sphere, respectively. Representing the In the first category channel, the... The initial basis state of 1 qubit.

5. The quantum evidence reasoning network pattern recognition method based on quantum evidence theory as described in claim 1, characterized in that, The initial quantum state is represented as a bit sequence consisting of multiple qubits, where the number of qubits in the bit sequence is . In each layer, parameterized single-bit rotation gates and controlled NOT gates are used alternately to extract correlation features within each channel to obtain multi-qubit features, including: A parameterized single-bit rotation gate is applied to each qubit corresponding to the initial quantum state; Apply controlled NOT gates sequentially in the forward direction of the bit sequence, such that the first bit... The number of qubits controls the first One quantum bit, ; Again, parameterized single-bit rotation gates are applied to each qubit; Apply the controlled NOT gate sequentially in reverse order along the bit sequence, such that the first bit... The number of qubits controls the first One quantum bit; A parameterized single-qubit rotation gate is applied to each qubit for the third time to complete the correlation feature extraction of the current layer, and the multi-qubit features are determined based on the extracted correlation features.

6. The quantum evidence reasoning network pattern recognition method based on quantum evidence theory as described in claim 4, characterized in that, The generation formula for the quantum fundamental probability amplitude function is as follows: ； in, Representing the zero state, on the Bloch sphere, Pointing in the positive direction of the Z-axis; Representing an excited state, on a Bloch sphere, Pointing in the negative direction of the Z-axis; Denotes the quantum fundamental probability amplitude function of the entire set. Indicates the first The basic probability amplitude function of a category; ground state The quantum fundamental probability amplitude function of the entire set is stored. ground state Stored the first Basic probability amplitude function of the category , Representative identification framework complete set.

7. A quantum evidence reasoning network pattern recognition system based on quantum evidence theory, used to implement the quantum evidence reasoning network pattern recognition method based on quantum evidence theory as described in any one of claims 1 to 6, characterized in that, The method is applied to pattern recognition tasks in image recognition, medical diagnosis, or physical signal classification, including: The acquisition module is used to acquire a classic feature dataset, which includes standardized classic features of multiple different data categories. The classic feature data is image data, medical detection data, or physical sensor data to be identified. The processing module is used to determine the number of qubits in each category channel based on the number of available qubits and the total number of data categories; allocate an independent category channel for each data category; map the normalized classical features to a rotation angle through an inverse trigonometric function; pass the rotation angle as a parameter to the rotation gate; and use the rotation gate to encode the classical data into an initial quantum state; the number of available qubits is the maximum number of physical qubits that the quantum computing hardware system executing this method can actually participate in quantum state encoding and circuit operations. The feature extraction module is used to input the initial quantum state into the variable quantum circuit. The variable quantum circuit contains a multi-layer structure. In each layer, parameterized single-bit rotation gates and controlled NOT gates are used alternately to extract the correlation features in each channel to obtain multi-qubit features. Then, the multi-qubit features are compressed to the last single bit through quantum pooling operation. An output rotation gate is applied to the last single bit to generate a quantum fundamental probability amplitude function carrying a complex phase. The fusion module is used to perform quantum tensor product operations on the quantum fundamental probability amplitude functions generated by each quantum channel and fuse them to form a composite quantum state that represents the global state. The output module measures the frequency of occurrence of a specific ground state in the composite quantum state and calculates the quantum fundamental probability distribution for each data category based on the following quantum evidence combination rule: ,in, The representative sample belongs to the channel category. The final quantum fundamental probability distribution, This indicates that the sample is divided into the first... A single category of a class. This indicates that the sample is divided into the first... A single category of a class; The total number of target categories in a pattern classification task; , , The value range is from 1 to ,in This is the category index where the target probability is currently being calculated. It is the category index used in the denominator for iterating and summing. This represents the index of auxiliary channels other than the target category; Representing the In the quantum fundamental probability amplitude function (QBPA) generated by each independent quantum channel, corresponding to the support class... The inherent trust level amplitude; Representing the In the quantum fundamental probability amplitude function generated by each quantum channel, corresponding to the entire set, The trust level amplitude; Represents the chain multiplication operator, where the specific ground state is only the first... The ground state is defined as having one qubit of 1 and the rest of the qubits of 0; and the data category with the highest quantum fundamental probability is selected as the identification result.

8. An electronic device, characterized in that, The electronic device includes: At least one processor (10); and, A memory (11) communicatively connected to the at least one processor (10); The memory (11) stores a computer program that can be executed by the at least one processor (10), which is executed by the at least one processor (10) to enable the at least one processor (10) to execute the quantum evidence reasoning network pattern recognition method based on quantum evidence theory as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program; when the computer program is executed by a processor, it implements the quantum evidence reasoning network pattern recognition method based on quantum evidence theory as described in any one of claims 1 to 6.

Images