33results about How to "Effective clustering" patented technology

The invention discloses a clustering method based on mobile object spatiotemporal information trajectory subsections. The clustering method based on mobile object spatiotemporal information trajectory subsections comprises the steps that the three attributes of time, speed and direction are introduced, and a similarity calculation formula of the time, speed and direction is provided for analyzing an internal structure and an external structure of a mobile object trajectory; firstly, according to the space density of the trajectory, the trajectory is divided into a plurality of trajectory subsections, then the similarities of the trajectory subsections are judged by calculating differences of the trajectory subjections on the space, time, speed and direction, finally, trajectory subsections in a non-significant cluster are deleted or are merged into adjacent significant clusters on the basis of a first cluster result, and therefore an overall moving rule is displayed on the clustering spatial form. According to the clustering method based on the mobile object spatiotemporal information trajectory subsections, the clustering result is improved, higher application value is provided, a space quadtree is adopted to conduct indexing on the trajectory subsections, clustering efficiency is greatly improved under the environment of a large-scale trajectory number set, and trajectories can be effectively clustered.

The invention relates to a vector-quantization-based overall and local color image searching method. The invention provides a new color image searching method, and relates to the field of an image processing technology. The method comprises the steps of converting an RGB (red, green and Blue) color space into an HSV (Hue, Saturation, Value) color space, performing relatively accurate clustering division on the color space by applying a neural-network-based competitive learning algorithm training code book; describing a space distribution situation of colors by introducing a color transfer matrix; combining the two characteristics, namely an index column diagram and a main color transfer matrix, so as to perform the similarity measurement; processing images by applying a morphological opening-and-closing operation, highlighting a target contour so as to extract a local interest area so as to highlight an important area and limit background information. The color image searching method overcomes the defects that the color space distribution description is not enough and the background information cannot be effectively limited by using the overall color histogram method. By using the vector-quantization-based overall and local color image searching method, the color quantization is relatively accurate, the matching effect is relatively good, and the vector-quantization-based overall and local color image searching method is an effective method for further improving the researching efficiency.

The invention relates to an image clustering method, which comprises the following steps of: creating a directional graph for a provided image sample set by using a variable bandwidth non-parameter nuclear density evaluation; partitioning the created directional graph into at least two non-intersected sub graphs by using a random walking isoperimetric partition method; and extracting image data in the sub graphs, and classifying the image data in the sub graphs into one category. The image clustering method fully considers the local probability density information of image data distribution, and can effectively cluster the data distributed extremely non-uniformly; and because the non-parameter clustering method is used, the method can process the image data with irregular shape distribution. Moreover, the invention also relates to an image clustering system.

Methods and apparatus, including systems and computer program products, to provide clustering of users in which users are each represented as a set of elements representing items, e.g., items selected by users using a system. In one aspect, a program operates to obtain a respective interest set for each of multiple users, each interest set representing items in which the respective user expressed interest; for each of the users, to determine k hash values of the respective interest set, wherein the i-th hash value is a minimum value under a corresponding i-th hash function; and to assign each of the multiple users to each of the respective k clusters established for the respective user, the i-th cluster being represented by the i-th hash value. The assignment of each of the users to k clusters is done without regard to the assignment of any of the other users to k clusters.

The invention relates to a personalized recommendation method based on dynamic neighboring point spectral clustering. A user-store bipartite network is set up according to user sign-in information; the user-store bipartite network is projected to a user-user one-aside network and a store-store one-aside network, and a node2vec algorithm is used for projecting the two weighted one-aside networks totwo different vector spaces; the spectral clustering algorithm based on dynamic neighboring points is called to cluster user vectors and store vectors obtained above to obtain multiple user clustersand multiple store clusters; the sign-in information existing between single users is converted into a cluster network among the user clusters and the store clusters; a K-means algorithm is used for dividing the one-dimension vector into two classes, the store clusters in the class with the larger sign-in average number are recommended to the user clusters; personalized recommendation is conductedaccording to each user cluster and the recommended store clusters. The accuracy of the recommendation method is improved effectively.

The invention particularly relates to a data clustering analysis method based on a genetic algorithm. The data clustering analysis method based on a genetic algorithm comprises the following steps: firstly, selecting an initial population from a sample set to be clustered; executing a genetic algorithm on the selected initial population; executing k-means operation on the new population generatedafter the genetic algorithm is executed; and repeating step (A) to step (C) until the optimal solution of the clustering problem is found. In the data clustering analysis method based on a genetic algorithm, local optimization of the K-means algorithm and global optimization of the genetic algorithm are combined, the optimal clustering number and the initial centroid set are finally obtained through multiple times of selection, intersection and variation genetic operation, the locality and sensitivity to the initial clustering center of a traditional K-means algorithm are overcome, and effective classification of data is achieved.

The invention discloses a user cluster analysis method in a customer care system. The method is characterized by establishing a systematic structure based on the customer care system, an application procedure of data mining and a brand-new algorithm. The method can overcome the defects in user clustering methods of current customer care systems, is scientific and proper, has high accuracy and strong universality, excellent effects, simplicity and reliability.

The invention discloses a stratospheric active routing design method based on a delay / disruption tolerant network. The active routing process is conducted in a manner of clustering, layering and cruising by unmanned aerial vehicles. The method comprises the following steps: firstly, dividing a user node into a plurality of mutually disjointed regions according to geographic positions through a clustering algorithm, and selecting an aggregation node of each region; secondly, sending the message to the aggregation node through the unmanned aerial vehicle in the region when a node in the region hopes to send a message to a user outside the region; thirdly, cruising the aggregation nodes through the unmanned aerial vehicle in the whole network; finally, uploading a message that a destination does not belong to the region to the unmanned aerial vehicle through the aggregation nodes when the unmanned aerial vehicle in the whole network passes through the aggregation nodes, forwarding the message to the aggregation node of the destination region thorugh the unmanned aerial vehicle in the whole network, and forwarding the message to the destination node through the unmanned aerial vehicle of the destination aggregation node.

The invention discloses a data clustering method for modal incomplete alignment. The method comprises: S1, acquiring a plurality of modal data sets, taking one modal data set as alignment modal data,and remaining analog non-alignment modal data; S2, respectively inputting each modal data set into one self-encoding network; S3, calculating a distance matrix of an alignment mode and a non-alignmentmode; S4, sending the distance matrix of the non-alignment modal data into a differentiable alignment module to calculate a prediction permutation matrix; S5, calculating a loss value by adopting a loss function; S6, performing back propagation based on the loss value to optimize the self-encoding network; S7, respectively inputting the modal data sets in the step S1 into the optimized self-encoding networks corresponding to the modal data sets; S8, obtaining a new prediction permutation matrix by adopting the execution modes of S3 and S4, and permuting the public representation output in thestep S7 by adopting the new prediction permutation matrix to obtain an aligned public representation; and S9, splicing the common representations output in the step S8, and then performing clusteringto obtain a clustering result.

The invention discloses a clustering method based on mobile object spatiotemporal information trajectory subsections. The clustering method based on mobile object spatiotemporal information trajectory subsections comprises the steps that the three attributes of time, speed and direction are introduced, and a similarity calculation formula of the time, speed and direction is provided for analyzing an internal structure and an external structure of a mobile object trajectory; firstly, according to the space density of the trajectory, the trajectory is divided into a plurality of trajectory subsections, then the similarities of the trajectory subsections are judged by calculating differences of the trajectory subjections on the space, time, speed and direction, finally, trajectory subsections in a non-significant cluster are deleted or are merged into adjacent significant clusters on the basis of a first cluster result, and therefore an overall moving rule is displayed on the clustering spatial form. According to the clustering method based on the mobile object spatiotemporal information trajectory subsections, the clustering result is improved, higher application value is provided, a space quadtree is adopted to conduct indexing on the trajectory subsections, clustering efficiency is greatly improved under the environment of a large-scale trajectory number set, and trajectories can be effectively clustered.

A methods and apparatus for distributing users to clusters are provided. In one aspect, a program operates to obtain a respective interest set for each of multiple users, each interest set representing items in which the respective user expressed interest; for each of the users, to determine k hash values of the respective interest set, wherein the i-th hash value is a minimum value under a corresponding i-th hash function; and to assign each of the multiple users to each of the respective k clusters established for the respective user, the i-th cluster being represented by the i-th hash value. The assignment of each of the users to k clusters is done without regard to the assignment of any of the other users to k clusters.

The invention discloses a point cloud misregistration filtering method and system for three-dimensional measurement of a complex special-shaped curved surface robot, and the method comprises the following steps: 1, dividing input registration point cloud pairs into three groups, selecting one point cloud pair from each group of registration point cloud pairs, and solving a corresponding transformation matrix Ri; 2, converting the obtained transformation matrix into an Euler rotation angle, and constructing an Euler rotation angle set D; Step 3, performing self-adaptive density clustering on the Euler rotation angles in the step D, selecting a class with the clustering result containing the maximum number of the Euler rotation angles, and taking a registration point cloud pair for solving the Euler rotation angles as a credible registration point cloud pair so as to complete error registration point cloud pair filtering; according to the method, spatial information of a given point cloud registration pair is fully utilized, effective decomposition and efficient clustering are carried out, so that false registration points in three-dimensional point cloud registration can be effectively eliminated, and the method has relatively high adaptability to noise of point cloud, registration position deviation and the like. The system is simple in structure and convenient to operate.

The invention discloses an infrared spectroscopy tea quality identification method mixed with GK clustering in a tea detection technology. A linear discriminant analysis method is used to learn a compressed training sample to acquire a training sample with identification information and a test sample with identification information. Fuzzy C average value clustering is carried out on the training sample with identification information to acquire the initial fuzzy membership degree and an initial clustering center. The fuzzy scattering matrix and the fuzzy membership degree value are calculated, and then a typical value is calculated. The clustering center is calculated according to the typical value. The Euclidean distance from the average value of the training sample with identification information to the clustering center of the test sample is calculated. If the Euclidean distance from the clustering center to the average value of training tea is the minimum, the tea variety of the clustering center and the tea variety of the training sample are the same. The tea and the category of the test sample are determined according to the fuzzy membership degree value. According to the invention, the typical value is added into a function, which can significantly reduce the probability of noise data processing errors.

The invention relates to a medical image clustering method, and belongs to the field of image processing. The method is based on a downhill algorithm, and comprises the following steps of preprocessing an image, extracting an image characteristic, searching a density attractor, searching hill foot pixels, and separating a layout image. Theories and practices prove that the method achieves a ridge, a flat top and a single threshold appropriately, and after the method is adopted, the medical image can be clustered efficiently, and the image characteristic is not lost, misled or omitted, so that the high-quality medical clustering image can be obtained, and diagnosis and reading requirements of medical workers are met.

The invention relates to the technical field of computer vision and point cloud segmentation, and provides a point cloud data set construction method and device based on statistics and concavity and convexity. A traditional point cloud segmentation method is applied to construction of a deep learning data set, and the problem that the deep learning point cloud data set is deficient is solved. According to the main scheme, the method comprises the following steps: step 1, obtaining target original point cloud data, and carrying out feature-based filtering and denoising; step 2, performing first clustering, and performing super-body clustering over-segmentation on the point cloud to obtain a voxel block set; step 3, clustering for the second time: performing LCCP clustering on each voxel block obtained in the step 2 to obtain an LCCP clustering set; step 4, third clustering: performing conditional Euclidean clustering on the LCCP clustering set based on the point feature histogram to obtain a final clustering set; and 5, marking the point cloud according to the result of the final clustering set, and organizing the file to obtain a point cloud data set.

According to the shadow rough fuzzy clustering method based on the Mahalanobis distance, the similarity measurement method of the Mahalanobis distance is adopted, correlation between attributes is eliminated, and the shadow rough fuzzy clustering method is suitable for any cluster division; through the Mahalanobis distance, the correlation between attributes is eliminated, and the difference of clustering importance of the attributes is reflected. A rough set and a shadow set are combined, the method is suitable for processing noise data and abnormal data, unbalanced sample distribution is improved, and the defect of a fuzzy C mean value in the aspect of ambiguity is overcome; meanwhile, according to the division of the core region and the boundary region in the cluster and the advantagesof the Mahalanobis distance, more effective cluster division can be generated.

The invention discloses a user load classification method based on a basic load reduction strategy, and the method comprises the steps: 1, extracting user load data, and carrying out the preprocessingof the load data; Step 2, selecting a load reduction reference value, taking the aggregation degree as an evaluation parameter, and distinguishing load types by adopting a method of reducing a reference load value; and step 3, clustering the load data by adopting a fuzzy C-means algorithm according to a load type distinguishing result. Aiming at the problem that the user type is mistakenly identified due to the user scale in the current load clustering method, the invention provides an improved load clustering algorithm processing skill, and the accuracy of the algorithm provided by the invention is verified by utilizing a fuzzy algorithm. The method adopted by the invention can effectively eliminate the interference of the user scale on the user type.

The invention relates to a data-field-based automatic clustering method. In the method, each space data is taken as a data point with mass, and the space data influence each other to form a data field, wherein total influence of all data points at a specific position represents the potential value of the data field. Points where the first-order partial derivative value of the potential value of the data field is zero are in an area where data superposition is most intensive, namely data cluster centers. The data-field-based automatic clustering method finds out the data cluster centers by searching the first-order derivative of the potential value of the data field, and searches the two sides according to the cluster centers and determines the edges of the clusters, thus finally marking complete clusters. Compared with the transitional clustering method, the data-field-based automatic clustering method has fast processing speed, is not influenced by noise, and is effective for clusters in any shapes, can be applied to the fields of image processing, community discovering, abnormal detection, market research and the like, thus improving the accuracy of the processing result.

The invention discloses a membrane calculation data cell clustering method oriented to the field of big data. through a k-means algorithm based on density, data is preprocessed, and the defect of manually reserving the number of the data clusters is overcome, and relatively reasonable data cluster division is obtained. A hybrid evolution mechanism adopts an Agnes algorithm based on hierarchical division, a genetic algorithm (GA) and a weighted fuzzy clustering (FCM) algorithm as evolution rules, and the advantages of the three clustering algorithms can be effectively combined in combination with membrane calculation, so that a better clustering result can be obtained.

The invention relates to a topic detection method based on document content and mutual relation. The method comprises the steps of preprocessing an obtained document to obtain a co-occurrence matrix and a two-two relation matrix of document features, constructing a target function on this basis, conducting iterative computation on a document representative degree matrix, a document subordination degree matrix, a word representative degree matrix and a word subordination degree matrix, and outputting the word representative degree matrix, wherein each row corresponds to one topic, and by using the word with the largest value in each row as a topic word for describing the topic, the topic word used for describing the topic is obtained. According to the topic detection method based on the document content and mutual relation, while document clustering and word clustering are conducted, joint clustering is more effective than respective clustering, a more comprehensive model is obtained when the relation of the document content and the relation among the documents are simultaneously considered compared with mere consideration of one information thereof, and through the introduction of the subordination degree and the representative degree, the method is not only applicable to clustering problems but also applicable to topic modeling problems.

The invention relates to the technical field of artificial intelligence, in particular to an intelligent prediction method for fusing fundus color pictures and deep learning myocardial infarction. Themethod comprises the specific steps of labelling and preprocessing collected fundus color pictures on the basis of longitudinal follow-up visiting data of a database; constructing a deep learning network, and optimizing deep parameters; obtaining patient fundus color pictures to be recognized, utilizing a deep learning intelligent model for recognizing fundus characteristics, and obtaining an intelligent classification result for judging whether or not myocardial infarction exists in the fundus color pictures; the method can efficiently, conveniently and non-invasively perform myocardial infarction high-risk group screening.

Embodiments of the present application provide an alarm processing method, apparatus, device, and computer-readable storage medium. The method includes: acquiring at least one time series data set; and determining at least one time series data set according to the at least one time series data set The identification of each abnormal time series and the abnormal start time of each abnormal time series; according to the identification and abnormal start time of each abnormal time series, the similarity between each abnormal time series is determined; according to the similarity between each abnormal time series, Determine the cluster ID of the alarm set; display the alarms in the alarm set according to the cluster ID of the alarm set. The method can accurately locate the abnormal start time and determine the similarity between multiple alarms, thereby effectively clustering multiple alarms and reducing the amount of alarms sent.

The invention belongs to the technical field of wireless communication, and discloses a user clustering method and system, a storage medium, a computer device and an application. The method comprisesthe steps: carrying out the initial user clustering according to the channel gains of all users in the system and the channel similarity between the users; calculating initial model parameters of theGaussian mixture model according to an initial user clustering result; calculating the posterior probability of each sample generated by each mixed component and iteratively updating the parameters ofthe Gaussian mixture model; and judging whether the number of iterations reaches a set termination condition or not: if so, performing user clustering according to a cluster mark, and otherwise, returning to calculate the posteriori probability of each sample generated by each mixed component and iteratively updating the parameters of the Gaussian mixture model. According to the method, the initial model parameters of the Gaussian mixture model are optimized, the system throughput is improved compared with a conventional user clustering method, and the method can be used for a multi-antenna non-orthogonal multiple access system.

The invention discloses a big data space-oriented data local density clustering method. The method comprises the following steps of presetting a set density parameter and a distance adjustment parameter; calculating a local density value of each data point; finding a maximum local density point in the data set in the calculation process; calculating a dynamic neighborhood radius to obtain a firstsub-cluster with direct density; similarly, obtaining each density data cluster of the original big data set; and according to the dynamic neighborhood radius and the adaptive density reachable distance of each attraction point, carrying out data set division. According to the method, the defects in the prior art are overcome, and effective clustering of clusters with different sizes, different forms and different densities is realized, so that help is provided for subsequent effective mining and analysis of big data.

The invention relates to a vector-quantization-based overall and local color image searching method. The invention provides a new color image searching method, and relates to the field of an image processing technology. The method comprises the steps of converting an RGB (red, green and Blue) color space into an HSV (Hue, Saturation, Value) color space, performing relatively accurate clustering division on the color space by applying a neural-network-based competitive learning algorithm training code book; describing a space distribution situation of colors by introducing a color transfer matrix; combining the two characteristics, namely an index column diagram and a main color transfer matrix, so as to perform the similarity measurement; processing images by applying a morphological opening-and-closing operation, highlighting a target contour so as to extract a local interest area so as to highlight an important area and limit background information. The color image searching method overcomes the defects that the color space distribution description is not enough and the background information cannot be effectively limited by using the overall color histogram method. By using the vector-quantization-based overall and local color image searching method, the color quantization is relatively accurate, the matching effect is relatively good, and the vector-quantization-based overall and local color image searching method is an effective method for further improving the researching efficiency.

The invention relates to a medical image clustering method, and belongs to the field of image processing. The method is based on a downhill algorithm, and comprises the following steps of preprocessing an image, extracting an image characteristic, searching a density attractor, searching hill foot pixels, and separating a layout image. Theories and practices prove that the method achieves a ridge, a flat top and a single threshold appropriately, and after the method is adopted, the medical image can be clustered efficiently, and the image characteristic is not lost, misled or omitted, so that the high-quality medical clustering image can be obtained, and diagnosis and reading requirements of medical workers are met.

The invention discloses a point cloud misregistration filtering method and system for three-dimensional measurement of a complex special-shaped curved surface robot, and the method comprises the following steps: 1, dividing input registration point cloud pairs into three groups, selecting one point cloud pair from each group of registration point cloud pairs, and solving a corresponding transformation matrix Ri; 2, converting the obtained transformation matrix into an Euler rotation angle, and constructing an Euler rotation angle set D; Step 3, performing self-adaptive density clustering on the Euler rotation angles in the step D, selecting a class with the clustering result containing the maximum number of the Euler rotation angles, and taking a registration point cloud pair for solving the Euler rotation angles as a credible registration point cloud pair so as to complete error registration point cloud pair filtering; according to the method, spatial information of a given point cloud registration pair is fully utilized, effective decomposition and efficient clustering are carried out, so that false registration points in three-dimensional point cloud registration can be effectively eliminated, and the method has relatively high adaptability to noise of point cloud, registration position deviation and the like. The system is simple in structure and convenient to operate.

Methods and apparatus, including systems and computer program products, to provide clustering of users in which users are each represented as a set of elements representing items, e.g., items selected by users using a system. In one aspect, a program operates to obtain a respective interest set for each of multiple users, each interest set representing items in which the respective user expressed interest; for each of the users, to determine k hash values of the respective interest set, wherein the i-th hash value is a minimum value under a corresponding i-th hash function; and to assign each of the multiple users to each of the respective k clusters established for the respective user, the i-th cluster being represented by the i-th hash value. The assignment of each of the users to k clusters is done without regard to the assignment of any of the other users to k clusters.