63 results about "Quantum neural network" patented technology

The invention provides a speaker recognition method combining a Gaussian mixture model and a quantum neural network. The method provided by the invention comprises the following steps: at the training stage, framing input training voice signals, extracting characteristic parameters and generating characteristic parameter vectors; then using a K mean value method and an EM (expectation-maximization) algorithm to obtain the Gaussian mixture model parameters of the characteristic parameter vectors, and finally utilizing the Gaussian mixture model parameters of all the training voice signals to train the quantum neural network; and at the recognition stage, obtaining the Gaussian mixture model parameters of a recognized speaker, then inputting the model parameters into the trained neural network, and obtaining a recognition result. The speaker recognition method is applicable to recognition of the speaker under the condition of less sample data and unbalanced sample data, and simultaneously the capability of the quantum neural network which can carries out effective judgment on voice data with cross data and fuzzy boundary of the speaker is utilized, so that the correct recognition rate of a system can be improved.

The invention provides a component mounting and dispatching optimization method for a chip mounter on basis of a quantum neural network. The component mounting and dispatching optimization method comprises the steps of establishing a mathematical model of the sum of paths required by the operation of mounting all components on a printed circuit board (PCB) according to an operating principle of mounting the components by a suction nozzle of the chip mounter, wherein the distances between all the mounted components and different feed tanks are taken as input vectors of the quantum neural network, and all weighting values are set as small random numbers, and given input vectors and target output vectors of a training set are provided; and calculating overall optimal solution of the established mathematical modeling by adopting a three-layer quantum neural network algorithm, thus obtaining an optimized component mounting and dispatching scheme corresponding with the sum of the shortest mounting patches of all the components, the optimal mounting order of all the components, and the arrangement positions of feeders of the components in the feed tanks. The component mounting and dispatching optimization method constructs the component mounting and dispatching mathematical model which takes both the component mounting order and the feeder arrangement positions into consideration, obtains the optimal control on the mounting and dispatching of the components in the PCB, shortens the mounting time of single-end arch components, and enhances the mounting and production efficiency of the components of the chip mounter.

The invention discloses a quantum data storage method and device, a quantum data reading method and device and computing equipment, and relates to the field of quantum computers. The quantum data storage method comprises the following steps: acting a first quantum neural network on an initial state to obtain an output quantum state; calculating a loss function based on a target quantum state and the quantum state output by the first quantum neural network, whereinthe dimension of the first quantum neural network is related to the dimension of the target quantum state, and the loss function corresponds to the distance between the target quantum state and the quantum state output by the first quantum neural network; adjusting parameters of the first quantum neural network according to the loss function so as to carry out iterative training on the first quantum neural network until a preset iterative stop condition is reached; and storing the trained parameters of the first quantum neuralnetwork in hardware equipment. According to the embodiment of the invention, storage of quantum data can be realized.

The invention provides an optimal multiuser detection method based on an evolutionary chaotic quantum neural network. The method comprises the following steps: establishing an optimal multiuser detection model; initializing an initial parameter of the chaotic quantum neural network, and activating the chaotic quantum neural network to acquire an approximate optimal solution; initializing the individual quantum, assigning the binary measurement state of the first individual quantum as the output value of the chaotic quantum neural network; constructing a fitness function and computing the fitness; evolving the quantum state of the individual quantum and acquiring a new measurement state by using a simulated quantum revolving door; activating the quantum neural network evolution mechanism in evolutionary chaotic scrambling to produce a sub-optimal solution for the binary state of each individual quantum; computing the fitness function value of each individual quantum to find out the global optimal solution; and outputting the global optimal solution as an optimal result for the multiuser detection. The detection method provided by the invention has excellent multi-access interference resistance and far-near effect resistance, wide application range, and can acquire the optimal detection result within the short time.

The invention discloses a quantum neural network-based comprehensive evaluation method for a multi-factor system. The method comprises the following steps of: 1) setting a plurality of feed-forward quantum nerve cells of the multi-factor system; 2) preparing a quantum register of n quantum bits; 3) taking l 0) or l 1) of the quantum register as the input of factor number of a user receiver, and taking l phi> as the output of the factor number, wherein l phi> is equal to the sum of costheta l 0) and sintheta l 1) ; 4) setting a parallel computing operator O which is used for the output quantum state of the comprehensive evaluation method for the multi-factor system and is updated and evolved; and 5) till the updated output quantum state and the change before update are in a permitted error range, namely the network state is stable, determining that a sending information sequence corresponding to the output quantum state is the detection result of the multi-factor system.

The invention discloses a binary classification method based on a quantum twin neural network. The method comprises the following steps: setting input data of a classification model; constructing a quantum neural network model, a quantum twin neural network model and a loss function model for training; learning and training the quantum twin neural network model according to the loss function modelto obtain a final quantum twin neural network classification model; and carrying out dichotomy on the to-be-classified data by adopting the quantum twin neural network classification model. The invention further discloses a face recognition method comprising the binary classification method based on the quantum twin neural network. According to the binary classification method based on the quantum twin neural network and the face recognition method of the binary classification method, the quantum twin neural network is adopted to carry out binary classification on the data, rapid binary classification of the data is achieved, and the method is simple, rapid, high in reliability and good in accuracy.

The invention provides a multi-fault intelligent diagnosing method for an artificial circuit utilizing quantum Hopfield neural network and aims at multi-fault coupling of the artificial circuit. The multi-fault intelligent diagnosing method includes the steps of data acquisition, feature extraction, feature quantization, fault cause probability analysis and the like. Ideal single-fault response and actual measured multi-fault response of the artificial circuit are obtained through SPICE simulation and a data acquisition board respectively. After wavelet packet decomposition, fault response wavelet coefficient defined by a new energy function realizes construction of an energy feature space. Elements in the energy feature space is submitted on the basis of quantization to a quantum Hopfield neural network model. Neuron states and connecting weight matrix in the network are expressed in quantum states. By calculating probability value of occurrence of related weight element in measurement matrix, the occurrence probability of quantum key input mode in forms of specific quantum memory prototype at specific time, and accordingly occurrence probability of multiple faults relative to specific single fault is obtained to judge fault types.

The invention discloses an encryption and decryption method based on a continuous variable quantum neural network. The method comprises the steps that the continuous variable quantum neural network isupdated; the sender and the continuous variable quantum neural network synchronously measure the basis; the continuous variable quantum neural network judges and preprocesses the plaintext sent by the sender and sends the plaintext back to the sender; the sender converts the preprocessed data information into a secondary plaintext on the basis of the synchronous measurement basis and sends the secondary plaintext to the continuous variable quantum neural network; the continuous variable quantum neural network encrypts the received information and sends the encrypted information to a receiver;and the receiver sends the encrypted information back to the continuous variable quantum neural network for decryption and obtains the information sent by the sender. By introducing a continuous variable quantum neural network model and a synchronous measurement technology, encryption and decryption of data are realized, and the method is high in reliability, good in security and easy to implement.

The invention provides a quantum neural network method and system for image recognition, and a medium, and relates to the technical field of quantum neural network methods, and the method comprises the steps: quantum state coding: preprocessing MNIST data set data, and converting the MNIST data set data into angle information corresponding to the operation of a rotating door according to a processing result; constructing a quantum neural network: optimizing the learning process of the quantum neural network through decomposition calculation; and characterizing a measurement result: finding out the quantum state with the maximum probability through the measurement result to realize image recognition. According to the invention, the structure logic of the quantum neural network is clearer, the realization is easy, and the learning efficiency is higher.

The invention provides a trend prediction method based on a depth quantum neural network, which comprises the following steps: 1. constructing an initial depth quantum neural network; 2, periodicallycollecting vibration signals of the bearing, and carrying out feature mining on the vibration signals by using wavelet packet decomposition; 3, training the depth quantum neural network model by usingthe train set data, and evaluating the performance of the model by using the verification set data; preprocessing the collected signals and dividing the processed feature parameters into training data set and testing data set. 4, adjusting the parameters of the depth quantum neural network model, and selecting an optimal prediction model for performance evaluation through continuously training the model; 5, predicting the trend of the bearing by using the prediction model. Through the above steps, the trained depth quantum gods network can realize the trend prediction of the bearing, and thebearing can be maintained in time through the trend prediction of the bearing, the repair time can be shortened, the repair cost can be reduced, and the mechanical failure problem caused by the bearing maintenance delay can be solved.

The invention belongs to the field of artificial intelligence, machine learning and quantum computing, and relates to a method for processing graph data through a quantum graph convolutional neural network. The method comprises preparing the preprocessed data into a plurality of quantum bits; constructing a quantum graph convolutional neural network model having a quantum bit input module, a quantum graph convolution module, a quantum pooling module, a quantum bit measurement module and a network optimization updating module; and iteratively training the model for multiple times and optimizing the parameters of quantum gates in the model, so that an output result reaches target output as much as possible, and a machine learning task is realized. According to the method, the non-Euclidean spatial data type machine learning task can be effectively processed by using the advantages of quantum calculation and the neural network, so that the quantum neural network is not limited to only processing structured data, and the application range of quantum machine learning is greatly expanded. In addition, the model is easy to package, has strong generalization performance, and can be expanded according to different graph data structures.

The invention discloses a handwritten picture classification method based on a quantum neural network. The implementation steps are as follows: (1) extracting handwritten picture features; (2) constructing a particle population of a binary quantum particle swarm algorithm; (3) constructing a convolutional neural network by using the particle population; (4) training the convolutional neural network; (5) selecting an optimal convolutional neural network; (6) judging whether the classification accuracy of the optimal convolutional neural network is smaller than 0.85 or not, and if yes, executingthe step (7); otherwise, executing the step (8); (7) updating the structure and parameters of the convolutional neural network corresponding to the position information of each particle by using a quantum updating strategy, and executing the step (3); and (8) outputting a classification result of the optimal convolutional neural network. The method has the advantages of being high in classification accuracy and capable of processing large-scale complex handwritten picture classification, and the problem that in the prior art, a large number of professional knowledge and design experiences ofthe convolutional neural network are needed is effectively solved.

A human body behavior recognition method based on the quantum neural network comprises the steps: 1, collecting human body action images, and carrying out graying on each frame of image; 2, storing the human body action image in a quantum state to form an NEQR quantum image; 3, quantum image foreground detection: respectively detecting a static target and a moving target for the NEQR quantum imageby adopting a background difference method and a ViBe algorithm; 4, using a Hadamard door edge detection algorithm to extract edge information of human body actions from the 'moving object' in the step 3 to form an edge quantum image; 5, performing convolution operation on the 'edge quantum image ' in the step 4 based on a convolution method of a quantum black box to extract feature points of actions, and making an identification sample set; and 6, reading the trained weight, and constructing a quantum BP neural network to identify the identification sample set. The scheme has the advantagesthat 1) quantum image processing is more flexible, 2) the parallel computing capability of quantum is fully utilized, and 3) the quantum neural network improves the accuracy of human body behavior recognition.

The invention provides a method for constructing a quantum feedforward neural network based on classical training. The method comprises the following steps: 1, giving a clear definition of quantum neurons; 2, selecting a specific activation function, and then using a quantum circuit for representing a model of quantum neurons; 3, proposing a quantum feedforward neural network model on the basis ofthe quantum neuron model in the step 2; and 4, providing a classic training method, quantitatively analyzing the effectiveness of the classic training method, completing the construction of the quantum feedforward neural network based on classic training. The problem that in the prior art, the clear definition of the quantum neural network is not unified is relieved through the method, and a quantum neural network model does not have the quantum states of input, output and weight at the same time; the realization of the activation function has no specific quantum circuit representation; the quantum neural network model has no ductility; and the quantum neural network lacks theoretical analysis on the effectiveness of the training process.

The invention discloses a distribution network voltage trend early warning method based on evidence theory fusion quantum network. The method comprises the following steps: S1, obtaining historical data of a distribution network; S2, preprocessing the historical data of the power distribution network, and dividing the historical data into a long-term (annual data) training set, a medium-term (seasonal data) training set, a short-term (week data) training set and a test set; S3, constructing an initial quantum neural network; S4, adjusting quantum neural network parameters, and training a quantum neural network model; S5, testing the neural network, and correcting a voltage trend prediction result in combination with the actual condition of the line; S6, improving the voltage trend prediction function of the quantum neural network through the DS evidence theory fusion; and S7, realizing a voltage abnormity early warning function. The invention can provide a new active prevention and control method for voltage abnormity of the distribution network and improve the power supply quality of the distribution network and the power grid operation economy.

Provided is a method for recognizing a signal mode of a cold-rolled strip. The method comprises the following steps that the plate shape measurement values, measured by a plate shape instrument on line, of all measurement sections in the width direction of the cold-rolled strip are acquired, and the plate shape values of all the measurement sections are obtained; original data output by the plate shape instrument are input into a n-layer neural network and used as a feature extraction layer, and the network is made to extract the feature automatically through training for eliminating the trace of artificial use; and plate shape recognition is conducted through the improved quantum neural network based on a genetic algorithm. According to the method, the improved quantum neural network of a multi-layer excitation function optimized through the genetic algorithm is applied to a plate shape mode recognition technology, the training efficiency of the network is improved remarkably, and the problems that the precision and the real-time performance are not ideal, the network structure is complex, the training time is long, and the stability and robustness are poor by means of a traditional plate shape recognition method are solved effectively.

The invention relates to a stock index price prediction method of a quantum neural network. The method is based on a main ensemble empirical mode decomposition algorithm, namely a PEEMD algorithm, and comprises a data input module, a data preprocessing module, a data conversion module, a data training and prediction module and a data reconstruction module. The data input module is used for acquiring latest transaction data of the stock index. The data preprocessing module is used for decomposing data, the data conversion module is used for converting original data into quantum state data, the data training and prediction module is used for carrying out training prediction on the quantum state data, and the data reconstruction module is used for reconstructing a prediction result of the data. The method comprises the following steps: firstly, preprocessing original data by using a PEEMD algorithm; decomposing non-stationary time sequence data into a plurality of approximate stationary data with different frequencies, removing high-frequency components in the data, only low and medium frequency components being subjected to simulation prediction through a quantum neural network, and finally reconstructing simulation results to obtain a final prediction result, so that the prediction performance of the model is effectively improved.

The invention discloses a quantum equipment calibration method and device, equipment and a medium. The quantum equipment calibration method comprises the steps of obtaining an actual output result of quantum equipment; inputting the actual output result into a quantum neural network for training to obtain a corresponding network output result; using the expected output result of the quantum device and the network output result to update the quantum neural network until the quantum neural network converges, and obtaining a trained quantum network line; and outputting a calculation result of the quantum device by using the trained quantum network line. Namely, equipment hardware is not directly debugged, training is carried out in a machine learning mode, the trained quantum network line is obtained, and the calculation result of the quantum equipment is output by utilizing the trained quantum network line, so that the quantum equipment can be effectively calibrated, the calibration complexity of the quantum equipment is reduced, and debugging is facilitated.

The invention provides a training method and device of a quantum state data processing model, electronic equipment and a medium, and relates to the technical field of quantum calculation, in particular to the technical field of quantum neural networks, and the specific implementation scheme is as follows: obtaining sample quantum state data, and determining an initial quantum state data processing model, the method comprises the steps of obtaining sample quantum state data, inputting the sample quantum state data into an initial quantum state data processing model to obtain output state data output by the initial quantum state data processing model, carrying out entanglement weighting processing on the output state data to obtain target output state data, and training the initial quantum state data processing model according to the target output state data to obtain target output state data. And obtaining a target quantum state data processing model. Therefore, when the target quantum state data processing model is adopted to execute the quantum analysis task, calculation resources occupied by quantum analysis can be effectively reduced, the quantum analysis efficiency and practicability are improved, and the quantum analysis effect is effectively assisted to be improved.

The invention discloses a dynamic modeling method for the combustion process of a circulating fluidized bed boiler. The dynamic modeling method comprises the following steps: the operation parametersof the combustion process of the boiler, which mainly affect the thermal efficiency of the boiler and the emission concentration of nitrogen oxides, are adjusted and recorded as input data and outputdata; Firstly, the input weights and the hidden layer thresholds of the sample incremental quantum neural network are determined according to the quantum computation rules. Then, based on the input data and the output data, the output layer weights and the weight matrix between the input layer and the output layer are calculated, i.e., the initialization model of the boiler thermal efficiency andNOx emission concentration is established. Based on the initialization model, the boiler operation parameters are collected on line, and the sample increment is calculated. The model parameters of thesample increment quantum neural network are updated in real time, including input weights and hidden layer thresholds, output weights and weights between input layer and output layer. Thus, the on-line models of thermal efficiency and NOx emission concentration are established, and the real-time modeling of boiler operating parameters is realized.

The invention relates to the technical field of heavy metal detection, in particular to a method for rapidly detecting heavy metal pollution to shellfish. The method comprises the following steps: firstly, preparing samples; secondly, carrying out hyperspectral image collection, correction, data extraction and preprocessing on the samples; thirdly, carrying out neighbourhood evidence decision making-based wave band selection on data, and extracting a subset of a characteristic waveband; fourthly, establishing a classification detection model, wherein the classification detection model comprises a quantum neural network classifier and an integrated learning classifier, the quantum neural network classifier is used for carrying out pollution and non-pollution detection classification on theshellfish by utilizing the subset of the selected waveband, and the integrated learning classifier is used for identifying and classifying different kinds of heavy metal pollution to the shellfish byutilizing the subset of the selected waveband; finally, obtaining a detection result of the samples. According to the method disclosed by the invention, data collection of the samples is carried out by utilizing a hyperspectral detection technology, waveband selection is carried out through the neighbourhood evidence decision making theory, classification detection is carried out by applying the quantum neural network classifier and the integrated learning classifier, the operation is simple and fast, better testing reproducibility is obtained, no any chemical reagent is required for assistingduring an analysis process, and pollution to environment is not generated.

Embodiments of a new approach for training a class of quantum neural networks called quantum Boltzmann machines are disclosed. in particular examples, methods for supervised training of a quantum Boltzmann machine are disclosed using an ensemble of quantum states that the Boltzmann machine is trained to replicate. Unlike existing approaches to Boltzmann training, example embodiments as disclosed herein allow for supervised training even in cases where only quantum examples are known (and not probabilities from quantum measurements of a set of states). Further, this approach does not require the use of approximations such as the Golden-Thompson inequality.

The invention discloses a brain wave recognition method based on a quantum neural network algorithm. The method comprises the following steps that step 1, a brain wave multichannel collection module collects brain wave signals; step 2, a brain wave multichannel signal amplification module amplifies the brain wave signals; step 3, a brain wave multichannel filtering module calls a quantum state multichannel preprocessing module and a quantum neural network module to filter the brain wave signals; step 4, a brain wave multichannel recognition module calls the quantum state multichannel preprocessing module and the quantum neural network module to carry out mode recognition and feature extraction on the brain wave signals; and step 5, the brain wave multichannel recognition module outputs a brain wave feature recognition result. According to the method, brain waves can be described by using a quantum state wave function, characteristics of coherence, superposition, wave particle bijection and the like among different brain wave signals can be described, and self-learning and training are further completed according to different scenes, so that the recognition effect is improved.

The invention provides a software vulnerability detection method and device based on a quantum neural network. The method comprises the following steps: step 1, positioning an API function in a target program to be detected; 2, slicing the target program to be detected according to the API function to obtain a plurality of code snippets; 3, standardizing a variable name and/or a function name in each code snippet; 4, constructing a dictionary based on the plurality of standardized code snippets, then coding each word in the dictionary according to a binary coding mode to obtain a binary vector corresponding to each word, and then performing quantum state angle coding on the binary vector corresponding to each word to obtain a quantum state corresponding to each word; and 5, inputting the quantum state corresponding to each word into the trained software vulnerability detection model based on the quantum neural network to obtain vulnerabilities in the target program to be detected.