18 results about "Tensor form" patented technology

The invention discloses a sparse polarization MIMO radar multi-parameter estimation method based on tensor decomposition, and the method comprises the steps: firstly constructing a dual-station MIMO radar system based on a polarized antenna, and calculating a covariance matrix of a receiving signal of a matched filter of a receiving end; then converting the covariance matrix into a tensor form, performing dimension transformation on the tensor, and obtaining a virtual covariance tensor of continuous virtual array elements in the difference joint array by using two transformation matrixes; dividing the virtual covariance tensor into a plurality of sub-tensors and connecting the sub-tensors along a fifth dimension and a sixth dimension; and finally, tensor decomposition is performed on the tensor obtained after connection, and multi-dimensional parameter estimation of the far-field target can be realized by using a matrix obtained through decomposition. The method has the advantages that joint estimation of the direction of arrival, the direction of departure and the polarization parameters is achieved under the underdetermined condition, and the estimation precision is high.

A method for decoding a plurality of tensors forming a hierarchical representation of feature maps for a single frame from a bitstream. The method comprises: decoding a first unit of information from the bitstream; decoding a second unit of information from the bitstream; and determining a first plurality of tensors, feature maps of at least one tensor of the first plurality of tensors having a different spatial resolution from feature maps of other tensor(s). The method also comprises determining a second plurality of tensors, feature maps of at least one tensor of the second plurality of tensors having a different spatial resolution from feature maps of other tensor(s). Feature maps of each tensor of the first plurality of tensors have different spatial resolution from feature maps of each tensor of the second plurality of tensors, and the tensors correspond to the hierarchical representation of feature maps for the single frame.

A system and method of encoding at least a plurality of tensors forming a hierarchical representation for a single frame into a bitstream. The method comprises deriving a first unit of information from a plurality of tensors forming the hierarchical representation, the plurality of tensors including at least a first tensor and a second tensor, feature maps of the first tensor having a larger spatial resolution than feature maps of the second tensor and encoding, in a first mode, at least the first unit of information into the bitstream. In a second mode, the method also comprises deriving a second unit of information derived from the first tensor; and encoding, the second unit of information and the first unit of information into the bitstream.

Various embodiments of the present disclosure provide systems and methods for generally generating and transforming physical quantum representations of multi-dimensional tensor data objects. Specifically, various embodiments enable the rapid and efficient generation of a physical quantum representation representing the Fourier transform of a multi-dimensional tensor data object based at least in part on manipulating another physical quantum representation of the multi-dimensional tensor data object itself via quantum manipulation operations. Information may be extracted from the generated physical quantum representation to determine the Fourier transform of the multi-dimensional tensor data object. Accordingly, various embodiments may comprise quantum manipulation operations for a tensor-form quantum Fourier transform (TQFT) for a multi-dimensional tensor data object. Various embodiments for the TQFT are advantageously comprehensive, versatile, and applicable to any quantum representation form for a multi-dimensional tensor data object. The TQFT may be performed in any quantum computing system and / or simulated quantum computing system.

This invention discloses an intelligent dynamic gesture recognition method, comprising: constructing initial node features of a dynamic gesture sequence obtained based on Leap Motion into a tensor form as subsequent input; designing a spatial feature extraction module based on a hypergraph neural network to learn the spatial features of the dynamic gesture sequence; designing a temporal feature extraction module based on a hypergraph neural network to learn the temporal features of the dynamic gesture sequence; constructing a spatiotemporal feature fusion classification module based on node spatial and temporal features to effectively fuse the spatiotemporal features of nodes and obtain the final gesture recognition result; and using a cross-entropy loss function to constrain network training based on the spatial feature extraction module, temporal feature extraction module, and spatiotemporal feature fusion classification module of the hypergraph neural network to obtain a dynamic gesture recognition model based on the hypergraph neural network. This invention utilizes the hypergraph neural network to effectively mine the potential higher-order correlations of gesture data, achieving more accurate dynamic gesture recognition.

The embodiment of the invention discloses an information prediction method and device, electronic equipment and a storage medium. The method comprises the steps that first information corresponding to a target object in multiple preset dimensions is acquired, and the first information of the target object in at least part of the multiple dimensions is empty; on the basis of a tensor form, representing the first information corresponding to the target object in each dimension to obtain tensor information; an information prediction model is obtained, and the information prediction model is obtained through pre-construction by combining a semi-tensor product and a link prediction model based on tensor decomposition; and inputting the tensor information into the information prediction model, and obtaining the first information of the target object in at least part of the dimensions according to an output result of the information prediction model. According to the technical scheme, consumption of computing resources can be reduced.

This invention presents a method for establishing large deformation viscoelastic constitutive relations based on Prony technology. It addresses the problems of existing viscoelastic constitutive models lacking universality, and the complexity and programming difficulties associated with their mathematical relationships. The invention includes: decomposing the stress tensor into the sum of the deviatoric stress tensor and the mean stress; decomposing the strain tensor into the sum of the deviatoric strain tensor and the mean strain; and providing a convolutional expression for the deviatoric stress tensor and the mean stress. The increment of the deviatoric stress tensor is expressed as a function of the deviatoric strain increment, and the increment of the mean stress is expressed as a function of the mean strain increment. The tensor form of the incremental constitutive relation is written in matrix form. This method has been programmed, and the numerical results obtained from the program agree well with the experimental results in specimen-level experiments on solid rocket motor propellant grains.

The present application belongs to the technical field of grain informatization processing, and particularly relates to a method and system for identifying imperfect grains based on spectral spatial coupling features. The method comprises inputting visible light images and hyperspectral data of the grains into a recognition model; a visible light image branch in the recognition model processes the visible light images of the grains to obtain the output of the branch; a hyperspectral data branch in the recognition model processes the hyperspectral data of the grains to obtain the output of the branch; the outputs of the visible light image branch and the hyperspectral data branch are used to obtain the recognition result; the recognition model is trained using grain sample images; the processing of the hyperspectral data branch comprises: expanding the hyperspectral data represented by a high-order tensor into a tensor factor multiplication mode containing spectral and spatial information features, and each mode is a tensor form in which all different spectral information features are superimposed; and the tensor form is subjected to feature extraction to obtain the branch output.

The application discloses a kind of intelligent substation merging unit fault diagnosis and self-healing method based on tensor decomposition, comprising: from the sensor network of intelligent substation, the operation data of merging unit is collected in real time, and is handled, and the data after processing is organized into tensor form, and output multi-dimensional data;Tensor decomposition method is used to decompose the multi-dimensional data, extract fault characteristics, and detect the fault characteristics, obtain detection result;Determine the position and type of fault based on the detection result, start self-healing control strategy, realize the automatic repair of fault by taking corresponding measures.The application realizes the automatic repair of fault by adaptive control technology, enhances the self-healing ability of intelligent substation, improves the reliability and stability of system.

The embodiment of the invention provides an abnormal account analysis method and device based on financial transaction, equipment and a medium. The method comprises the following steps: in response to a financial transaction request of a user, obtaining to-be-transacted data information and a behavior portrait model of the user; wherein the behavior portrait model of the user represents historical transaction behavior characteristics of the user in at least three dimensions; tensor chain feature information of the to-be-transacted data information is determined, and behavior feature information is determined according to the tensor chain feature information; wherein the tensor chain feature information represents to-be-transacted data information in a tensor form, and the behavior feature information represents a behavior attribute of a user indicated by the to-be-transacted data information; determining state information of the to-be-transacted data information according to the behavior portrait model and the behavior feature information of the user; wherein the state information represents the condition that the account of the user indicated by the to-be-transacted data information is abnormal. The method is used for improving the efficiency and precision of analyzing the abnormal account based on the financial transaction.

An apparatus and method for encoding a plurality of tensors forming a hierarchical representation for a single frame into a bitstream. The method comprises producing a first tensor from the plurality of tensors and producing a second tensor from the first tensor using at least three blocks, each of the three blocks comprising a convolutional layer, and encoding the second tensor into the bitstream. An output tensor of a convolutional layer of a first block among the three blocks has a smaller channel count and a smaller spatial size compared to an input tensor of the convolutional layer of the first block, an output tensor of a convolutional layer of a second block among the three blocks has a smaller channel count and a smaller spatial size compared to an input tensor of the convolutional layer of the second block, an output tensor of a convolutional layer of a third block among the three blocks has at least one of a smaller channel count and a smaller spatial size compared to an input tensor of the convolutional layer of the third block.

3999397_2 Abstract METHOD, APPARATUS AND SYSTEM FOR ENCODING AND DECODING A TENSOR An apparatus and method for decoding at least a plurality of tensors forming a hierarchical representation of feature maps for a single frame from a bitstream. The method comprises: decoding a first unit of information from the bitstream; decoding a second unit of information from the bitstream; and determining a first plurality of tensors from the first unit of information, feature maps of at least one tensor of the first plurality of tensors having a different spatial resolution from feature maps of other tensor(s) of the first plurality of tensors. The method also comprises determining a second plurality of tensors from the second unit of information, feature maps of at least one tensor of the second plurality of tensors having a different spatial resolution from feature maps of other tensor(s) of the second plurality of tensors. Feature maps of each tensor of the first plurality of tensors have different spatial resolution from feature maps of each tensor of the second plurality of tensors, and the tensors of the first plurality of tensors and the second plurality of tensors correspond to the hierarchical representation of feature maps for the single frame Abstract 20 26 21 40 59 07 A ug 2 02 6 A b s t r a c t 2 0 2 6 2 1 4 0 5 9 0 7 2 0 2 6 A u g

The application discloses an incomplete multi-view clustering method fusing consensus and difference information, and belongs to the technical field of multi-view clustering methods. The method firstly constructs an incomplete graph by using the cross-order neighborhood relation of non-missing samples, so as to mine potential comprehensive and complementary neighborhood information in data. Subsequently, a logarithm-based tensor rank approximation model is introduced to perform low-rank recovery on a tensor formed by the incomplete graph, so that a complete fine-grained affinity tensor is obtained. On this basis, the application further proposes an adaptive weight fusion mechanism, which learns the weights of different views to extract a consistent affinity matrix from the recovered tensor. The mechanism can fully utilize the difference and consensus of cross-views, and effectively solves the clustering problem caused by the missing of multi-view data.

The invention relates to the technical field of motor adaptive control, and discloses a motor adaptive control method and system based on artificial intelligence, and the method comprises the steps: collecting multi-source operation data in a motor operation process; constructing a tensor representation with at least a three-order structure based on the collected multi-source operation data; carrying out high-dimensional feature compression, and extracting a low-dimensional hidden space feature vector representing the running state of the motor; constructing a control objective function, and calculating the optimal control quantity of the motor; driving a motor to execute control operation based on the calculated optimal control quantity; on the basis of the feedback data, carrying out online updating on a tensor auto-encoder model; the system comprises a data acquisition module, a tensor construction module, a tensor auto-encoder module, a control optimization module, a control execution module and an online updating module. The method introduces a tensor form with a three-order structure to express the multivariable operation data of the motor, and can capture the coupling characteristics of a time sequence, a space and variables at the same time.

The invention provides a community search method and system oriented to new hardware. The method comprises the following steps: acquiring an original image to be subjected to community search and converting the original image into a tensor form; constructing an index based on-truss hierarchical decomposition for the original image in the tensor form through a tensor operator; and performing community search on the index based on the-truss hierarchical decomposition by a community search algorithm based on tensor calculation to obtain a community search result of the original image. According to the method, the original image is converted into the tensor format, and the efficient index based on-truss hierarchical decomposition is constructed, so that efficient community search based on new hardware is realized.

A data processing method and device, electronic equipment and storage medium. The data processing method comprises: obtaining a first original tensor form of first to-be-processed data, the first to-be-processed data being stored in a storage unit; obtaining a mapping relationship between the first original tensor and a first target tensor, wherein the granularity of a single operation of the first original tensor is a plurality of tensor elements; calculating index information of the first target tensor according to the mapping relationship between the first target tensor and the first original tensor; and performing a transformation operation on the first original tensor according to the calculated index information of the first target tensor, so as to transform the first original tensor into the first target tensor. The data processing method can reduce the number of tensor transformations, reduce resource consumption, and improve system performance.

The embodiment of the invention discloses a distributed training method based on a database, and the method employs a load balancer in a distributed database system as a parameter server in distributed training, and is used for storing model parameters of a to-be-trained model. The method comprises the steps that a to-be-trained model is built in a distributed database system, distributed training is carried out in cooperation with a plurality of database servers in the distributed database system, the database servers can recognize and analyze data stored in a tensor form, and input, operation and output operations of at least part of the to-be-trained model are completed by running UDF contained in SQL statements and utilizing locally stored tensor data. According to the method, the multiple devices in the distributed database system are directly utilized to perform distributed training on the to-be-trained model, data does not need to be exported from the database, the data does not need to be converted into a tensor form, and the model training efficiency can be effectively improved.

The application provides a traffic prediction method based on low-rank tensor dynamic mode decomposition under data missing condition. First, a tensor form of dynamic mode decomposition is introduced, and dynamic changes of each time sequence of a day are expressed as a dynamic tensor composed of different state transition matrices, so as to capture dynamic characteristics in traffic data. Then, a mask operator is introduced, and errors between reconstructed observation tensors and original observation tensors with missing data in non-missing data parts are as small as possible. After that, considering time periodicity and spatial similarity of traffic data, global low-rank constraints and time constraints are applied to the dynamic tensor. Finally, the reconstructed observation tensor is solved to obtain a prediction result. Experiments prove that under the influence of data missing, the method provided by the application can effectively realize a traffic prediction task.