258 results about "Distance matrix" patented technology

The invention relates to a hydrological flow prediction method and system based on multi-site time-space association. The prediction method comprises the following steps: carrying out dimension reduction and feature reconstruction on original multi-site hydrological data through an auto-encoder; space-time correlation modeling and adjacency matrix dynamic construction are carried out, a multi-dimensional Euclidean distance matrix between stations is calculated based on a multivariable dynamic time warping (MDTW) algorithm, a similarity matrix is generated in combination with dynamic programming, and a dynamic adjacency matrix is constructed by fusing a geographic space adjacency relation; extracting spatial features of a GCN (Graphics Convolutional Network); carrying out adaptive time sequence decomposition and trend-period modeling; and carrying out multi-stage fusion prediction and result output, and generating a final prediction result through a decoder in combination with the decomposed trend item and periodic item. According to the method, accurate extraction and dynamic correlation modeling of spatial-temporal characteristics of multi-site hydrological data are realized, the accuracy and robustness of single-site flow prediction are improved, and the problems that multi-site spatial-temporal correlation modeling is insufficient, non-linear time sequence alignment is difficult, and single-site prediction precision is limited are solved.

The invention provides an electric energy quality device power smooth flexible capacity expansion method for a new energy access scene. The method comprises the following steps: acquiring three-phase voltage, current and power signals, obtaining standardized three-dimensional tensor data through multi-dimensional tensor spectrum projection decomposition, calculating a characteristic value and a characteristic tensor base, and generating a dynamic characteristic vector and a weight coefficient set; a multi-layer probability graph matrix is constructed, a feature mapping matrix and a hierarchical contribution coefficient are obtained through iterative calculation, a fluctuation feature tensor is constructed, and a multi-time scale prediction model is established; constructing a control manifold space and a measurement tensor, calculating a topological connection matrix, designing a multi-dimensional control law and performing stability optimization; and constructing an embedded feature matrix and an inter-module distance matrix, calculating a capacity expansion weight and a power distribution scheme, and performing spectral domain equalization control and grid-connected coordination control. According to the invention, the power smoothing effect is obviously improved, and the system capacity expansion flexibility is enhanced.

The invention relates to the technical field of image analysis, in particular to a dispensing track optimization control system based on visual template conversion, which comprises a visual template analysis module, a contour point position correction module, a track vector generation module, a multi-parameter linkage feedback module and an optimal path planning module. According to the method, a boundary is extracted through clustering pixel superposition brightness and channel weight, a line segment frequency screening track is counted, feature stability and precision are enhanced through datum line construction, a deviation value is calculated through gradient direction segmentation boundary, a matching degree is improved through tangent point interpolation correction contour, and a weight factor is generated through included angle change and tool parameter normalization. A reference point is optimized through superposition increment, a path fusion tool attribute dynamic adaptation parameter is adopted, a multi-dimensional data fusion visual offset and floating weight real-time adjustment track is acquired, an execution point and an adjustment vector are superposed to calculate a distance matrix, an optimal path is screened through minimum distance and energy consumption, efficiency and consumption are both considered, and the system robustness and execution economical efficiency are remarkably improved.

The invention discloses a data dynamic partition storage method and system based on adaptive clustering, and relates to the field of data processing. The method comprises the following steps: S1, extracting multi-scale geometric features of a high-dimensional data set, calculating a local curvature and generating a curvature feature matrix; s2, constructing a feature distance matrix and a similar matrix based on the curvature feature matrix, and generating a low-dimensional embedding matrix; s3, executing self-optimization clustering according to the low-dimensional embedded matrix, determining a cluster number through singular value distribution, and generating an initial cluster; s4, calculating a stability factor of each cluster, and triggering cluster splitting or merging operation according to a threshold value to form updated cluster division; and S5, performing incremental processing on newly added data points, obtaining a new point local curvature through local curvature gradient correction, mapping the new point local curvature to a low-dimensional space, dynamically deciding affiliation based on a cluster radius, and updating partitions. Through multi-scale feature extraction, self-optimization clustering and incremental updating mechanisms, high-dimensional data clustering precision, robustness and calculation efficiency are improved.

The invention discloses a power load curve clustering method, system and equipment based on an improved density peak value clustering algorithm and a storage medium, and belongs to the field of intelligent power distribution networks, and the method comprises the steps: carrying out the sampling segmentation of power load data, constructing a daily load curve set, carrying out the standardization processing of each daily load curve, and generating a standardized sample set; based on the standardized sample set, calculating a sample distance relation of the daily load curve set, and constructing a corresponding distance matrix; establishing a local density estimation model according to the distance matrix, and calculating a density estimation value of each sample curve; according to the local density estimation value of each sample curve and the relative distance between each sample curve and other samples in the distance matrix, constructing a clustering decision value set, and selecting a plurality of sample curves with clustering decision values higher than a threshold value as clustering centers; and by taking the clustering center as a reference, sequentially distributing the residual sample curves to the class cluster to which the corresponding clustering center belongs according to the density estimation value and the distance relationship between the residual sample curves and each clustering center, and completing sample clustering division. According to the method, the problem of misclassification when the density difference of the power load curve set is too large is solved, the sample density difference when the data density difference is too large is accurately represented, the clustering precision of the load curve is effectively improved, reliable power consumer social attribute identification is provided for a power supply side, and an energy planning strategy is better implemented.

The invention relates to the technical field of acceleration signal recognition, in particular to an acceleration signal recognition method and device, and the method comprises the following steps: carrying out the first-order numerical differential operation of a three-axis acceleration signal based on a collected original three-axis acceleration signal, and obtaining a movement speed sequence of the signal; according to the method, first-order and second-order numerical differential operation is carried out on an original three-axis acceleration signal, and the speed change and the impact change of the signal are extracted at the same time and combined to form a kinematic feature set, so that the dynamic features of the motion state can be fully expressed in the two dimensions of speed and impact. The Euclidean distance between the to-be-identified signal and the motion template is calculated based on the kinematic feature set to form the basic distance matrix, and the constraint cost matrix is constructed in combination with the speed sequence and the impact signal, so that the constraint on the kinematic consistency in the signal comparison process is enhanced, and the matching path is more reasonable.

The invention discloses a data optimization storage method and system for smart agriculture, and relates to the technical field of smart agriculture, and the method comprises the steps: collecting agricultural field data, remote sensing data and management data, and carrying out the preprocessing of the data, and constructing a multi-source heterogeneous data matrix; the method comprises the steps of calculating fuzzy similarity of a multi-source heterogeneous data matrix, constructing a time sequence distance matrix, calculating weight vectors of different data types, constructing a weighted distance matrix according to distribution weights of different data types, calculating a weighted fuzzy similarity matrix, constructing a Boolean similarity matrix, distinguishing redundant nodes and representative nodes, and determining difference vectors. According to the method, redundant data are accurately identified by using the weighted fuzzy similarity matrix, only the difference vector is stored, the storage space occupation is greatly reduced, the consumption of cloud or chain end storage resources is effectively reduced, the actually transmitted data volume is obviously reduced through storage compression and differential storage, and the data transmission efficiency is improved. And the data transmission burden of the edge device or the cloud node is reduced.

The invention relates to the technical field of point cloud semantic segmentation, and discloses a point cloud segmentation method and system based on point cloud serialization and a Mama network. Specifically, a geometric distance matrix and a semantic distance matrix are calculated for a plurality of representative point units obtained by sampling on an original point cloud, and a path sequence is generated based on a fused comprehensive distance matrix. And traversing the path sequence by adopting a bidirectional scanning strategy to generate a forward sequence and a reverse sequence, extracting features from the forward sequence and the reverse sequence by utilizing a Mama network, cascading or fusing the extracted features, and finally performing classification based on the fused features. According to the method, better long-range dependency and local context modeling effects can be obtained, the comprehensive expression capability of detail features and global context in sparse and irregular scenes is enhanced, and the calculation and memory overhead required for global dependency capture of large-scale point clouds is greatly reduced.

The invention provides a computing power sensing method and system with integrated computing network. The method comprises the following steps: constructing and analyzing a multi-level NUMA hierarchical structure diagram to obtain an inter-node distance matrix and a resource distribution condition; obtaining an optimized resource mapping scheme based on the inter-node distance matrix and the resource distribution condition; obtaining a node subset passing verification based on the resource mapping scheme and the information of the plurality of computing nodes; performing asynchronous Byzantine consensus processing by adopting a directed acyclic graph structure based on the request of the node subset passing the verification to obtain a consensus result of consistency of nodes of the whole network; based on the consensus result and the data feature distribution information of the plurality of computing nodes, obtaining a model knowledge transmission result between adjacent nodes; and based on a model knowledge transmission result, obtaining an optimization result of an end, edge and cloud three-layer model. According to the invention, a unified perception mechanism of end-edge-cloud cooperation is realized, the system response speed in a distributed environment is improved, and the computing power prediction accuracy is improved.

The invention discloses a multi-target path optimization method, equipment and a medium, and relates to the technical field of path planning. The method comprises the following steps: constructing a weighted graph model of a to-be-planned area and verifying an adjacent matrix to obtain verified weighted graph data; shortest path distances among all node pairs are calculated to form an all-source shortest path distance matrix and a precursor record set, and connectivity detection and feasible region pruning processing are carried out on unreachable node pairs; generating an optimized access sequence covering all target nodes based on the processed full-source shortest path metric data and a nearest neighbor greedy strategy, accumulating composite path cost, and performing feasibility check of time windows, capacity and risk constraints to generate an initial path sequence and accumulated cost data; and expanding adjacent node pairs in the initial path sequence into a specific executable path on the original graph according to the precursor record set, performing neighborhood optimization on the executable path by adopting a local search operator, and outputting an optimized final path scheme and an accumulated cost report.

The invention provides a global structure sensing vector quantization method based on optimal transmission, which comprises the following steps of: mapping input data to a continuous hidden variable space through an encoder to generate continuous representation; constructing a distance matrix between the continuous representation and the codebook, and solving a distribution matrix by using an optimal transmission problem; carrying out normalization processing on the distance matrix, and generating an initial value of a distribution matrix by adopting a special initialization method; performing row and column normalization iteration on the distribution matrix through a Sinkhorn-Knopp algorithm until the distribution matrix is converged; and after iteration is finished, quantitative characteristics are determined based on the maximum value of the distribution matrix, the quantitative characteristics are mapped back to an original data space through a decoder, and a reconstructed data result is obtained. Based on the scheme provided by the invention, a codebook utilization rate close to 100% and an excellent data reconstruction result can be ensured, and a powerful tool is provided for discrete compression scene reconstruction of continuous data such as images and videos.

The invention discloses a perception-assisted unmanned aerial vehicle track and power distribution joint optimization method and system, and the method comprises the steps: firstly, extracting environment related features from high-dimensional CSI data, constructing a weighted graph, and calculating a shortest path distance matrix; and the CSI is mapped to a two-dimensional space through a twin neural network to maintain distance consistency, and user distribution is obtained by combining with known position correction. On the basis, an optimization model taking the minimization of the energy consumption of the unmanned aerial vehicle as a target and the user rate demand as a constraint is established, the optimization model is decomposed into a power distribution and trajectory optimization sub-problem, and the Lyapunov optimization and successive convex approximation are used for alternate solution. According to the method, high-precision positioning is realized through the channel map under the scene that the user position is unknown, the energy consumption of the unmanned aerial vehicle system is reduced by 27% or above through joint optimization, and meanwhile, 98% of user communication requirements are guaranteed.

The invention discloses a short-term power prediction method and system based on a wind power plant cluster. The method comprises the following steps: constructing a similarity distance matrix of a key meteorological element time-varying rule between different stations; calculating a Copula entropy between the historical power samples of each station, and constructing a spatial similarity matrix of station power according to each Copula entropy; performing normalization processing on the similarity distance matrix and the spatial similarity matrix, and summing the similarity distance matrix and the spatial similarity matrix after normalization processing to obtain a space-time similarity distance matrix; clustering the similarity distance matrix by adopting a hierarchical clustering algorithm, determining a partitioning scheme in combination with the geographic distribution of the new energy stations, and dividing each wind power plant in the target region into each region; and outputting a total power prediction result of the obtained region according to the LSTM time sequence prediction model. The overall precision and robustness of power prediction are improved, and a more reliable scheduling basis is provided for large-scale access of a new energy station to a power grid.

The invention discloses a gesture recognition method, system and device based on fine-grained clustering and a medium, and the method comprises the steps: inputting an obtained gesture image into a teacher model, extracting a high-dimensional feature embedding vector of a specific layer, decomposing the high-dimensional feature embedding vector into a plurality of fine-grained sub-blocks, carrying out the clustering of the sub-blocks through a distance matrix between the sub-blocks, and carrying out the recognition of the fine-grained sub-blocks. According to the similarity between each clustering set and the original gesture image, distributing a weight for each clustering set, and carrying out dot multiplication on the weights and the label matrix to generate a weight matrix for subsequent knowledge distillation; the student model performs weighted distillation learning by utilizing the teacher features extracted by the teacher model and the weight matrix through a dynamic weight regulation and control mechanism; through a hierarchical feature distillation mechanism, automatically extracting fine-grained features of gestures from deep features and shallow features of the teacher model and the student model through clustering, and accurately describing the features of the gestures; knowledge of a complex model is migrated to a lightweight model, and efficient operation is ensured.

The invention provides a multi-feature factor matching and interrupt processing method for storm tracking. The method comprises the following steps: calculating storm features required by tracking; carrying out discretization processing and coding on the storm features, and constructing a sparse matrix; constructing an auto-encoder model, performing data reconstruction on the sparse matrix based on the auto-encoder model, and calculating main features of the single storm; on the basis of the main features, calculating the feature vector distance between the previous and later time secondary storms, and constructing a feature space distance matrix; a bipartite graph algorithm is adopted, the incidence relation of the previous and later time-order storms is obtained based on the feature space distance matrix, and a storm matching result is constructed; identifying an interrupted storm based on a storm matching result and carrying out secondary matching; and constructing a storm trajectory based on a secondary matching result and removing a repeated storm trajectory. According to the method, the movement track of the storm can be accurately tracked from formation to extinction of the storm, reliable data support is provided for meteorological monitoring and forecasting, and the timeliness and accuracy of severe convective weather early warning are improved.

The invention relates to the technical field of power grid analysis, and discloses a power grid topological structure extraction method, a power grid fault prediction method and a computer storage medium, and the method comprises the steps: firstly obtaining the data of each node in a power grid, constructing a power incidence matrix through the obtained data, determining the edge weight between the nodes through the power incidence matrix, and obtaining the edge weight of each node; constructing a vertical edge weight matrix, and establishing a diagonal angle matrix by using the edge weight matrix; constructing a weighting matrix based on the edge weight matrix and the diagonal angle matrix; and finally, constructing an objective function of a weighting matrix based on the power injection amount, solving the objective function through a gradient descent method to obtain an optimized weighting matrix, constructing a distance matrix by using the optimized weighting matrix, clustering each node by using the distance matrix, and extracting a power grid topological structure according to a clustering result. According to the extracted power grid topological structure, the accuracy of the connection relation between the nodes is improved, and the precision of the power grid topological structure is improved.

A system and method for detecting anomalies in a data stream is described. The system receives the data stream that comprises values of metrics derived from observations of operation of a computing entity over a time window. A model comprising variances of the data over the time window is formed. The model identifies operating thresholds for each metric based on the variances of the data for each metric in the data stream. The system computes a steady state distance matrix of the data stream. The system determines that the steady state distance matrix exceeds a steady state threshold. In response to determining that the steady state distance matrix exceeds the steady state threshold, the system computes a pattern distance matrix based on the steady state distance matrix. The anomaly in the data stream is detected based on the pattern distance matrix. The system generates an alert indicating the anomaly.

The invention belongs to the technical field related to data processing, and provides a brain network analysis method and system based on multi-network collaborative topology analysis in order to solve the problem that an analysis result is unstable due to threshold selection subjectivity in existing Alzheimer disease brain network analysis. Extracting brain regions corresponding to the default mode network, the significance network and the execution control network, calculating correlation coefficients among different brain regions to construct a function connection matrix, and converting the function connection matrix into a distance matrix meeting complex construction requirements; constructing a Vietories-Rips complex based on the distance matrix, calculating a coherence group under each scale of the filtering sequence, and extracting topological features of a 0-dimensional Betti number and a 1-dimensional Betti number; according to the method, the Betti number is extracted, a curve that the Betti number changes along with the threshold value is drawn, the difference of the two groups in topological characteristics is compared, the specific topological biomarker related to the Alzheimer's disease is identified, and richer biomarker information is provided for early diagnosis of the Alzheimer's disease.

The invention relates to the technical field of automatic driving, and provides a denoising optimization method and device for three-dimensional point cloud data, equipment, a medium and a product, and the denoising optimization method for the three-dimensional point cloud data comprises the steps: carrying out the fusion processing of a color image and the point cloud data, and obtaining a preliminary point cloud data set; performing simplification processing on the initial point cloud data set to obtain a target point cloud data set; in the target point cloud data set, constructing a graph structure according to the color information and the geometric structure, and determining weights of edges in the graph structure; and establishing a noise reduction model according to the graph structure, carrying out iteration on the noise reduction model, and alternately updating the feature distance matrix and the estimated value of the noise reduction point cloud until a convergence state is reached, so as to obtain denoised point cloud data. According to the method, multi-source data fusion, point cloud simplification, graph signal and graph structure construction and an iterative optimization algorithm are applied, point cloud data polluted by noise are efficiently processed and recovered, noise can be effectively suppressed, and key details and features in the point cloud data can be accurately reserved.

The invention relates to the technical field of urban road traffic state analysis and processing based on the Internet of Things, and discloses an urban road traffic state analysis and processing method based on the Internet of Things. A complete analysis chain from sensor data acquisition, error calibration and clock synchronization, road section length extraction by a high-precision electronic map, vehicle density calculation, graph structure modeling, distance matrix generation, attenuation parameter calculation, spectrum matrix construction and characteristic value solving, density prediction, anomaly detection, signal lamp adjustment and path optimization pushing is constructed. Through graph topology modeling and graph calculation, urban traffic is abstracted into a graph model with spatial characteristics and structure association, and the traffic density diffusion modeling and analysis capability is remarkably improved. A power iteration and residual error detection strategy is adopted to replace an empirical rule, so that a self-adaptive anomaly detection and signal regulation and control mechanism is realized, and the real-time performance and the intelligence of the system are enhanced.

The invention relates to an industrial pollution discharge equipment operation data analysis method and system, and the method comprises the steps: carrying out the normalization of each index of collected industrial pollution discharge equipment operation data, calculating the distance matrix of each index, obtaining a minimum spanning tree through the distance matrixes, calculating the distance average value in the minimum spanning tree of each index, and obtaining a minimum spanning tree of each index; setting a weight for each index according to the distance average value; calculating the distance between the operation data according to the weight, obtaining the local density and the high-density minimum distance of the operation data according to the truncation distance and the distance, determining the number of nearest neighbors according to the minimum spanning tree of the index, calculating the operation data and the standard deviation of the local density of the nearest neighbors of the number of the nearest neighbors of the operation data, and obtaining the standard deviation of the local density of the nearest neighbors of the operation data. Determining a cluster center by using the local density, the high-density minimum distance and the standard deviation; and clustering the industrial data according to the cluster center, and displaying a clustering result as a data analysis result in a chart form.

The invention discloses a taxi route recommendation method based on a FastDTW time sequence clustering algorithm. The method comprises the following steps: firstly, acquiring original GPS track data of a taxi from a public data platform, and preprocessing the data; thirdly, calculating the similarity between the trajectories by using a FastDTW algorithm, and constructing a distance matrix between every two trajectories; then, the distance matrix is applied to a hierarchical clustering algorithm, a high-frequency destination hot spot area is mined, and a high-probability passenger-carrying destination is obtained; and finally, in combination with factors such as the driving distance, the driving time and the driving cost, calling an Amod map # imgabs0 # to obtain detailed route information, and calculating and recommending an optimal route through comprehensive scores. According to the invention, based on hot destination prediction and multi-target route optimization of data mining, the optimal passenger searching route is recommended for the unloaded taxi, the unloaded rate of the taxi is reduced, the passenger loading probability of the unloaded taxi is improved, and time and resource consumption are reduced, so that the vehicle scheduling efficiency and the driver operation efficiency are improved.

The invention discloses a translation memory bank retrieval and matching method based on semantic similarity, which relates to the technical field of computer-aided translation, and comprises the following steps: generating an internal edge and a cross-graph edge, calculating a one-time jump success probability, calculating edge and cross-graph edge cost based on cleanliness, correcting by using the one-time jump success probability to obtain an edge weight; generating a weighted graph, constructing an initial priority queue, searching by using a shortest path to obtain a distance matrix, calculating density based on a fuzzy similarity matrix, screening the density to obtain a genealogy center, calculating proximity, and generating candidate translations in combination with the distance matrix; static and dynamic fragment sets are constructed through a dependency syntax and BERT-NER, semantic matching precision and noise robustness are improved, a weighted graph is constructed in combination with cleanliness and transfer factors, shortest path matching is carried out, and the retrieval recall rate and quality of technical translation are improved.

The invention relates to a modeling method for accurate traffic flow prediction, which comprises the following steps of S1, integrally modeling all intersections and roads into a road network undirected graph, and defining an adjacent matrix and a distance matrix, S2, defining propagation time delay of traffic flow at the intersections, and S3, calculating the traffic flow of the intersections according to the propagation time delay. S3, combining propagation time delay and traffic flow abrupt change influence of adjacent intersections to confirm a comprehensive effect of each intersection at the moment t, and obtaining traffic flow representation of the intersection at the moment t, S3, aggregating features of each intersection through an adjacent matrix and the propagation time delay by a space graph convolutional layer, and capturing a dynamic rule through a time graph convolutional layer, the method comprises the steps of (S1) obtaining a graph convolution layer, combining the graph convolution layer and a time convolution layer to form a space-time graph convolution network, and outputting final traffic flow prediction, (S2) giving the final traffic flow prediction and a kernel density estimation matrix of GKDE and setting three learnable matrixes, and (S5) outputting through a feedforward neural network. The method has the advantage of accurate traffic flow prediction.

The present disclosure relates to a method, apparatus, device, storage medium, and program product for text similarity measurement. The method includes: obtaining a first text string and a second text string; constructing a joint probability distribution of the two based on the first text string and the second text string, and sampling the joint probability distribution to obtain a sampled string; calculating the distance from the first text string to the sampled string to obtain a first distance matrix, and calculating the distance from the second text string to the sampled string to obtain a second distance matrix; determining the similarity between the first text string and the second text string based on the first distance matrix and the second distance matrix. The embodiments of the present disclosure achieve dimensionality reduction calculation of the text similarity method, improving the calculation efficiency. In addition, since the sampled string contains information of the two text strings, information loss is reduced while dimensionality reduction is achieved.

The invention belongs to an SDTW-IPAM-based short-term power load prediction method and system, electronic equipment and a storage medium. The method comprises the steps of data preprocessing, clustering analysis, single-class prediction model construction and future load prediction. The clustering analysis comprises setting of upper and lower limits of a cluster number, construction of a distance matrix, calculation of a Gap value and a standard error, selection of an optimal cluster number, generation of an initial center, clustering processing and normalization processing; according to the method, a distance measurement method is introduced into load clustering, so that the dynamic time sequence characteristics of a load curve can be described more accurately, the local time deformation of the load curve can be effectively identified, and the distinguishing capability of similar load days under the influence of weather, events or potential new energy fluctuation is improved; a clustering algorithm is improved by using statistics and an initialization strategy, so that the stability and the accuracy of performing mode recognition on complex load data are improved, and the load can be effectively divided into typical modes with different dynamic characteristics.

The invention relates to a gearbox fault diagnosis model construction method based on metric guide graph comparative learning, and belongs to the field of gearbox fault diagnosis model construction. The method comprises the following four core stages: firstly, carrying out frequency domain conversion and normalization on vibration signals of the gearbox to generate a node characteristic matrix; secondly, a cosine distance and an Euclidean distance are fused to construct a mixed distance matrix, and a fault diagnosis graph is generated based on a K-nearest neighbor algorithm; then unsupervised graph comparison pre-training is realized through graph data enhancement and a dynamic graph attention network (DGAT); and finally, weak supervision fine tuning is carried out by using a small number of marked samples to complete construction of the gearbox fault diagnosis model. According to the method, the construction of the high-precision gearbox fault diagnosis model can be realized in a scene with extremely few marked samples (1-10 samples per class), and the method is suitable for planetary gearbox health monitoring in the fields of wind turbines, helicopters, hybrid electric vehicles and the like.

The invention relates to the technical field of safety protection, can be applied to business system platforms of financial science and technology, medical health and the like, and discloses a Byzantine attack defense method, device, equipment and medium based on federated learning. Carrying out forward propagation on the sample data set by utilizing a local model, constructing a plurality of probability matrixes, determining the Euclidean distance between any two probability matrixes, constructing a distance matrix, carrying out hierarchical clustering on target clients to obtain a plurality of client clusters, obtaining a probability standard matrix of a standard model, and obtaining the probability standard matrix of the standard model; and determining the similarity between the probability matrix and the probability standard matrix one by one, performing type marking on the to-be-marked matrix according to the similarity, performing information updating on the client cluster to obtain an updated cluster, and aggregating all the updated clusters to obtain a final defense model. According to the method, the defense efficiency and the defense scene coverage rate of the Byzantine attack can be effectively improved.

Systems and methods for slide image alignment are described herein. An example method includes receiving a plurality of slide images, detecting a plurality of features contained in the slide images, and comparing a plurality of pairs of the slide images. The comparison uses the detected features. The method also includes creating a distance matrix that reflects a respective difference between each of the pairs of the slide images, creating a graph by connecting each of the slide images to its most similar slide image, and detecting a plurality of graph components. Each of the graph components includes one or more of the slide images. The method further includes aligning the slide images within each of the graph components, and aligning the graph components to form a composite image.

The invention provides an MDS dimension reduction-based retrieval enhancement generation semantic matching optimization method and system and a storage medium. The method comprises the following steps of: obtaining a user problem and high-dimensional semantic vectors of a plurality of candidate text segments; constructing a distance matrix based on the two high-dimensional semantic vectors to obtain a high-dimensional semantic similarity; performing dimension reduction on the distance matrix by adopting a multi-dimensional scale analysis (MDS) method, and calculating low-dimensional semantic similarity between the user question and each text fragment based on a distance relationship after dimension reduction; performing weighted fusion on the low-dimensional semantic similarity and the high-dimensional semantic similarity to generate a mixed similarity; and sorting the mixed similarities from high to low, and selecting the text fragments corresponding to the top n of the mixed similarities for generating response output. By reducing the dimension of the high-dimensional retrieval vector to the low-dimensional space, the interference of irrelevant dimensions is reduced, the semantic matching precision and efficiency are improved, and then the quality of the text recalled by the large language model is improved.