211 results about "Hypergraph" patented technology

The present invention relates to a method of intelligent 2D and 3D object and scene modeling, transformation and manipulation and more particularly this invention relates to the field of computer modeling, virtual reality, animation and 3D Web streaming. The method uses attributed hypergraph representations (AHR) for modeling, transforming and manipulating objects. From one or more 2D views of a 3D object or scene, range information is first computed and then a triangular mesh model is constructed. The data structure is designed to handle the transformations on the representation corresponding to movements and deformations of the object. In an attributed hypergraph, the attributes associated with the hyperedges and the vertices facilitates modeling of various shapes with geometrical, physical or behavior features. As a hierarchical and generic representation, AHR enables pattern matching, recognition, synthesis and manipulation to be carried out at different resolution levels on different subsets depending on the context. Symbolic computation on knowledge represented in the format of attributed hypergraphs becomes straightforward. Given the features of a 3D object or scene, the procedure of constructing the AHR corresponds to the concept of functor in category theory, which maps one category to another one. The transformations of AHR are in the form of a set of operations defined on attributed hypergraphs, which stand for the motions and deformations of the object. This representation is applied to various modeling and manipulation tasks on 3D objects. The process of motion analysis of a 3D object is the task of extracting a sequence of AH operators from the AHR of the object. A 3D scene can be modeled by AHR and then altered / augmented with other 3D models, by which an augmented reality can be built. Given the AHR's of two different 3D shapes, 3D morphing may be accomplished by matching the two AHR's and then mapping the difference to a sequence of AH operators. Model based animation of an object can be accomplished by applying a set of AH operators to its AHR. The AHR method forms a data compression system for efficient web streaming over the Internet.

Apparatus, systems, and methods may operate to create a hypergraph of weighted vertices comprising computing resources and storage resources, and nets comprising workloads; to receive a plurality of requests to be addressed by a network associated with the hypergraph, at least some of the requests associated with data objects; to calculate partition schemes for the network based on the requests and the data objects according to an ant colony optimization heuristic; and to autonomously reallocate the workloads to the computing resources and / or the storage resources according to the partition schemes. The workloads may act as ants following a path defined by the vertices of the hypergraph. Further activities may thus include depositing pheromones along hyperedges of the hypergraph, wherein the hyperedges are used for swapping the vertices between the workloads. Additional apparatus, systems, and methods are disclosed.

The invention belongs to the technical field of cross-media relevance learning, and particularly relates to a social image oriented cross-modality image-label relevance learning method. The invention comprises three algorithms: multi-modal feature fusion, bidirectional relevancy measuring and cross-modality relevancy fusion; a whole social image set is described by taking a hyperimage as a basic model, the image and a label are respectively mapped into hyperimage nodes for treatment, relevancy aiming at the image and the relevancy aiming at the label are obtained, and the two different relevancies are combined according to a cross-modality fusion method to obtain a better relevancy. Compared with the traditional method, the method is high in accuracy and high in adaptivity. The method has important significance in performing efficient social image retrieval by considering multi-modal semantic information based on large-scale social images with weak labels, retrieval relevancy can be improved, user experience is enhanced, and the method has application value in the field of cross-media information retrieval.

The invention provides a method for automatically screening influential users on social media networks. The method comprises the following steps of: modeling for users, interested objects and interaction relationship between the users and the interested objects by a hypergraph model; automatically learning to obtain a connotative interest topic by using a regularization topic probability model of hypergraph restraints and taking the similarity relationship between the content information and the content information of the interested objects as restraints; and performing topic influence sorting on each user and interest object, propagating topic influence by using a similarity propagation model, the users and the interest objects on the hypergraph and the hyperedge between the users and the interest objects until to reaching a steady state, and then sorting to obtain the influential users under specific topics. By the method, the distribution of the user influence in the social media network can be truly and accurately reflected.

The invention provides a hypergraph convolutional network model and a semi-supervised classification method thereof. The method comprises the following steps: 1, carrying out sparse coding on sample data features of a non-Euclidean structure to form a sample representation coefficient matrix; 2, constructing a hyperedge according to the similarity of the samples, calculating a hyperedge weight, and constructing a hyper-graph model; 3, by means of a hypergraph theory, defining convolution operation on a hypergraph, and constructing a hypergraph convolution network model; 4, defining a semi-supervised learning method on the hypergraph convolutional network, designing a loss function, and predicting category labels of all samples by using category information of a small number of calibrationsamples; 5, respectively making a semi-given label matrix for training, verifying and testing, setting network hyper-parameters, training a network model, and learning a convolution kernel and a regularity factor parameter of the network according to a random gradient descent algorithm; and 6, for given data, predicting an unknown sample category by using the trained model to realize semi-supervised classification.

Disclosed herein is a method of processing images based on image segmentation using higher-order correlation clustering. In an image segmentation method according to an embodiment of the present invention, an input image is segmented into superpixels. A hypergraph is constructed by connecting two or more adjacent superpixels, among the superpixels, to one another. A joint feature map is created by extracting feature vectors from respective edges of the hypergraph, and partitioning the hypergraph based on higher-order correlation clustering in consideration of specific constraints.

The invention discloses an RGBD image salience detection method based on a hypergraph model. The method includes conducting hyperpixel segmentation for a color image to be detected and a depth image, calculating neighborhood depth contrast graphs for each hyperpixel area of the depth image, constructing a depth background hyperedge according to neighborhood depth contrast ratio, extracting hyper pixel areas on the boundary of the image to construct a boundary background hyperedge, calculating the weight of the two hyperedges, expanding hyperedges according to a hypergraph learning algorithm, building an induce graph, prior calculating boundary background salient map by using the boundary connectivity based on the spatial adjacent relation of the induce graph and the edge weight, and obtaining a final salient detection map based on salience degree updating algorithm of cellular automaton and a fusion algorithm in combination with depth prior. Deficiency of conventional 2D boundary background prior is overcome. The improvement is made based on depth information and a hypergraph model, and better effects are achieved than a conventional image salience detection method that combines color and depth information.

A symbolic disjunctive image computation method for software models which exploits a number of characteristics unique to software models. More particularly, and according to our inventive method, the entire software model is decomposed into a disjunctive set of submodules and a separate set of transition relations are constructed. An image / reachability analysis is performed wherein an original image computation is divided into a set of image computation steps that may be performed on individual submodules, independently from any others. Advantageously, our inventive method exploits variable locality during the decomposition of the original model into the submodules. By formulating this decomposition as a multi-way hypergraph partition problem, we advantageously produce a small set of submodules while simultaneously minimizing the number of live variable in each individual submodule. Our inventive method produces a set of disjunctive transition relations directly from the software model, without producing a conjunctive transition relation—as is necessary in the prior art. In addition, our inventive method exploits the exclusive use of live variables in addition to novel search strategies which provide still further benefit to our method.

The invention provides a multi-task personalized web service method based on a hypergraph, comprising the following steps: relations among objects in a social network community are acquired, the relations among the objects are constructed to form a hypergraph, each object is constructed to be a node in the hypergraph, each relation among the objects is constructed to be a hyperedge in the hypergraph, the objects of the same type form nodes of the same type, the relations among the objects of the same type form the hyperedges of the same type in the hypergraph, and each hyperedge is weighted; a matrix H for characterizing an inclusion relation between the hyperedges and the nodes and a weight diagonal matrix W are acquired; the matrix H and the weight diagonal matrix W are subject to precomputation, query objects are selected in accordance with personalized service demands, and query vectors are constructed in accordance with the query objects; the other objects on the hypergraph are sequenced in accordance with correlation between the others objects and the inquiry objects; and screening is carried out on sequencing results in accordance with the personalized service demands, and the result is pushed to the user. The multi-task personalized web service method based on the hypergraph has the advantages that various types of information in the network community can be taken into account simultaneously, multidimensional relation can be processed, and non-loss of multidimensional information can be ensured.

Systems, methods, and computer storage media for handling failures in generating structured queries from natural language queries. One of the methods includes obtaining, through a natural language front end, a natural language query from a user; converting the natural language query into structured operations to be performed on structured application programming interfaces (APIs) of a knowledge base, comprising: parsing the natural language query, analyzing the parsed query to determine dependencies, performing lexical resolution, forming a concept tree based on the dependencies and lexical resolution; analyzing the concept tree to generate a hypergraph, generate virtual query based on the hypergraph, and processing the virtual query to generate one or more structured operations; performing the one or more structured operations on the structured APIs of the knowledge base; and returning search results matching the natural language query to the user.

The invention discloses a method and a device for fusing multiple machine translation systems, relates to the related field of machine translation, and aims at solving the problems that the information of a decoding process is not fully considered by the traditional method for carrying out system fusion on post-treatment and a search space in decoding cannot be fully considered. The device for fusing of the multiple machine translation systems comprises a preprocessor, a phrase extractor, a voice model generator, a plurality of machine translation system trainers and a decoder. The method comprises 1, pretreating machine translation systems; 2, building a translation hypergraph of each translation system; 3, fusing two translation hypergraphs and training a training set, wherein training comprises two parts, wherein a single machine translation system before fusion adopts a BTG ordering model of the maximum entropy training, and the machine translation system after fusion adopts MERT of the minimum error rate training; 4, decoding a test set to generate a translation result, and grading the translation result. The method and the device are applied to the field of machine translation.

The invention relates to a method for image co-segmentation based on a hypergraph. The method comprises the following steps that two images with the similar or the same foregrounds are read in; segmentation process is conducted on the input images through the Mean-shift algorithm; statistical operation is conducted on a color histogram H of the images through a histogram statistical method; the similarity of adjacent sub blocks in one image and the similarity of similar sub blocks in the different images are calculated; a similarity matrix of nodes and a similarity matrix of hyperedges are calculated, the proper value and the proper vector of the Laplacian matrix of the hypergraph are calculated and the foregrounds and backgrounds are segmented. According to the method, calculating speed of segmentation is improved, the foregrounds are well related, the backgrounds are well related, more relative information of the sub blocks is kept, and the segmenting effect is improved.

The invention provides a distributed-structure-based big data clustering method. The distributed-structure-based big data clustering method comprises the following steps: S100, preprocessing big data; S200, segmenting and managing the big data; S300, establishing a clustering hypergraph model; S400, mapping the big data, specifically, respectively mapping segmented data blocks to the hypergraphs H=(V, E), namely mapping each data block to one hypergraph; S500, clustering all the data blocks respectively by using the hypergraphs; S600, reclustering clustering results of all the data blocks, obtained in the step S500, to obtain a final clustering result. According to the distributed-structure-based big data clustering method, by using a cloud platform in combination with a hypergraph theory, the big data is mined and clustered, so that fast, real-time and accurate analysis and processing of the big data are achieved.

The invention discloses a cross-modal retrieval algorithm based on mixed hypergraph learning in subspace, based on the cross-modal public subspace learning of canonical correlation analysis. The similarities inside and among modals are calculated by a mapping of the public subspace. A mixed relation matrix is calculated by the similarities inside and among the modals. A mixed hypergraph model is built by the extraction of the relation matrix. And the cross-modal retrieval and sample sequencing are conducted by adoption of hypergraph learning finally. With aiming at cross-modal heterogeneous variations and high order relation among samples, the algorithm instance applies the hypergraph model combined with the cross-modal public subspace learning to the cross-modal retrieval, so that the model is capable of considering the similarity among the modals and the similarity inside the modals simultaneously and giving consideration to the high order relation among a plurality of the samples meanwhile, improving the final precision ratio and the final recall ratio of the cross-modal retrieval. The algorithm is capable of effectively improving the performance of the cross-modal retrieval and greatly enhancing the precision ratio and the recall ratio of the cross-modal retrieval.

A method (and structure) to retrieve data from information structures based on an entity / relation paradigm and characterized as being a self-similar hypergraph, includes creating a template that matches a self-similar hypergraph format of the information structure. The template contains at least one query unit.

The embodiment of the invention provides a method and a system for image segmentation. The method comprises the following steps: pre-segmenting a given image to obtain a plurality of sub-areas; merging the plurality of sub-areas according to at least two features to obtain sub-segmentation result corresponding to each feature; indicating sub-segmentation results corresponding to at least two features as hypergraphs; and performing cluster ensemble on the sub-segmentation results of the at the least two features based on the hypergraphs to obtain the segmentation result for the given image. The method is effectively combined with segmentation for the image based on multiple features, overcomes the defect that a common segmentation algorithm has no generality based on single image feature and avoids the problem that the features have no due effect on segmentation performance due to improper setting of visual feature vector and partial weights and dimensions in Bag of words.

The invention relates to a hierarchical FPGA layout and wiring method based on a multi-level method and empowerment hypergraph. A mathematical model of a logical netlist in a hierarchical FPGA layout and wiring process is constructed by using empowerment hypergraph; circuit logic units are represented as nodes in the empowerment hypergraph; wiring among the circuit units is expressed as overrides in the empowerment hypergraph. The multi-level method is adopted as a hierarchical FPGA layout and wiring method to effectively meet the higher requirements of VLSI on the operating efficiency and processing capacity. The method can gradually reduce the scale of problems and the time complexity of methods by bottom up cluster stage and also can process details of problems level by level, increasing solution accuracy of methods by means of top-down projection optimization so as to be naturally combined with hierarchical features of researched objects. By means of the hierarchical FPGA layout and wiring method based on the multi-level method and empowerment hypergraph, the time complexity of FPGA physical design can be effectively reduced and the design capability and performance of FPGA are increased.

The invention provides a rail transit passenger flow prediction method based on a dynamic hypergraph convolutional network, relates to the field of deep learning and the like, and particularly relatesto a hypergraph representation and graph convolutional network-oriented flow prediction task. According to the method, on the basis of carrying out high-order representation on the topological relation of the rail transit network by utilizing a hypergraph, the introduction of a graph convolution network is realized through a hypergraph convolution module, and a dynamic hypergraph convolution network mechanism is realized by excavating the internal space-time characteristics of passenger flow OD to construct a dynamic hyperedge. Compared with a traditional mathematical model and a traditionalmachine learning method, the method is deeper and more accurate in rail transit feature extraction. The task of rail transit passenger flow prediction is completed, and the prediction accuracy is improved.

The invention provides a microblog emotion prediction method based on multi-mode hypergraph learning and belongs to the field of multi-mode emotion analysis. The microblog emotion prediction method based on multi-mode hypergraph learning is provided aiming at the problems existing in emotion prediction on microblog multi-channel content. The method includes the following steps that 1, microblog multi-mode features are extracted; 2, the distance between microblog is calculated; 3, a multi-mode hypergraph model is established; 4, hypergraph learning is carried out. Different modes can be better associated to solve the problem of independence between the modes, and a good effect is achieved on microblog emotion prediction.

The invention discloses a hypergraph spectrum analysis based image segmentation method which specifically includes the steps of image preprocessing, image superpixel uniting, hypergraph model generation and hypergraph spectrum clustering. Areas after image over-segmentation are used as superpixels which are taken as vertexes, relation among the superpixels is obtained via various different over-segmentation methods to construct a hyperedge so as to form a hyperedge model, a traditional method for establishing relation (edge) between every two pixels to form a graph model is replaced, and accordingly high-order coupling relation among the pixels is reflected effectively, relevance among the areas of an image is considered, and area consistency and edge accuracy of a segmentation result are enabled to be obviously improved.

The invention provides a three-dimensional object retrieval method and device based on hypergraphs, wherein the method comprises the following steps: computing a distance matrix among all views of a three-dimensional object in a database; clustering all the views according to the distance matrix to obtain a plurality of clustering results, and constructing a plurality of hypergraphs corresponding to the three-dimensional object according to the clustering results; fusing the hypergraphs to form a fused hypergraph, analyzing the fused hypergraph, and establishing the relationship between the fused hypergraph and the three-dimensional object according to the analytic result; and retrieving the three-dimensional object according to the relationship. By using the method provided by the invention, the problem of low retrieval accuracy caused by complex three-dimensional object information is well solved. By establishing a model through hypergraphs, the method can effectively analyze the relationship of the three-dimensional object, thereby obtaining more accurate and effective retrieval effect.

Systems, methods, and computer storage media for handling failures in generating structured queries from natural language queries. One of the methods includes obtaining, through a natural language front end, a natural language query from a user; converting the natural language query into structured operations to be performed on structured application programming interfaces (APIs) of a knowledge base, comprising: parsing the natural language query, analyzing the parsed query to determine dependencies, performing lexical resolution, forming a concept tree based on the dependencies and lexical resolution; analyzing the concept tree to generate a hypergraph, generate virtual query based on the hypergraph, and processing the virtual query to generate one or more structured operations; performing the one or more structured operations on the structured APIs of the knowledge base; and returning search results matching the natural language query to the user.

Decoupling capacitors (dcaps) are placed in an IC design by assigning different dcap utilization rates to logic cones, applying the rates to corresponding dcap regions surrounding cells in the cones, identifying any overlap of regions from different logic cones, and inserting a dcap at the overlapping region having the highest dcap utilization rate. The best location for the dcap is computed using a hypergraph wherein the cells are edges and the regions are nodes. Any node that is dominated by another node is removed and its edge is extended to the dominating node. The dcap is inserted in the region having the most edges (the edges can be weighted). The process is repeated iteratively, updating the hypergraph by removing nodes connected to dcap location, and inserting the next dcap at a region corresponding to the node which then has the greatest number of connected edges.

The invention discloses a search method for relevance feedback images based on an ant colony algorithm and a probability hypergraph, which comprises a training stage and a search stage. The training stage comprises the following steps of extracting low-level features of images, studying a dictionary and performing the high level representation of the images in a library image. The search stage comprises the following steps of extracting low-level features of sample images; performing the high level representation of the sample images; constructing an affinity matrix; initializing or updating a pheromone matrix; labeling positive correlation images and negative collation images for search results of all sample images in the image library; calculating a semantic pheromone matrix; calculating an affinity enhancing probability; enhancing the affinity matrix by using the ant colony algorithm; constructing a hypergraph; and returning the search result and finishing search, or updating the pheromone matrix for the next search. According to the method, an efficient and accurate image search technology is provided for image search and a higher use value is realized.

The invention discloses a multi-target tracking method. The multi-target tracking method comprises the steps that S1, a video is divided into time quanta in the time domain, target detection is carried out on each video frame, and the detection results serve as suspected target areas; S2, the positions of the suspected target areas obtained in the S1 serve as vertexes, otherwise, local tracks in the time quanta serve as vertexes, the relation between the local tracks serves as edges, and an affinity graph of the local tracks is built; S3, a plurality of intensive neighbor kinds are obtained from the affinity graph in a searching mode, and the local tracks belonging to the same intensive neighbor kind serve as local tracks belonging to the same target; S4, unit time quanta belonging to the same time quantum are aggregated to form a plurality of new unit time quanta, and local tracks, belonging to the same intensive neighbor kind, in the same new unit time quantum are merged to form a merging track; S5, whether a single time quantum has crossed the whole tracking time domain or not is judged, if yes, the merging track obtained at present serves as a tracking track of a monitored target, and if not, the S2 is executed.

The invention discloses an image set unsupervised co-segmentation method based on deformable graph structure representation. The method is divided into the following four steps: firstly, superpixel segmentation is carried out on each graph in a graph set and a descriptor is extracted; secondly, an inner graph based on a Biharmonic distance is built for each picture, specifically, a two-dimensional image is stretched as a three-dimensional grid, and a Laplacian matrix is established on the three-dimensional grid according to a Riemann flow pattern space manner; thirdly, a deformable hypergraph which includes all superpixel in the graph set is established after the inner graph is established and the Biharmonic distance is calculated in a descriptor space(characteristic space); fourthly, clustering ideas considering connectivity are used, and an energy function which comprises the inner graph, the hypergraph and segmentation size constraints is established. Optimization is carried out by using the expectation maximization algorithm to obtain a final co-segmentation result. The algorithm partly uses CPU(Central Processor Unit) parallelism with CPU parallelism, the co-segmentation method has excellent segmentation accuracy and efficiency performance for the large-scale graph set.

The invention discloses a social media graph representation model-based social risk event extraction method. The method comprises the following steps of 1) modeling an event by adopting an HCCG model, defining an entity relationship generation rule, describing event attributes, and performing multi-granularity extraction on the event by utilizing word-level and stream-level contexts; 2) performing similarity calculation by utilizing an information quantity ratio of a maximum common subgraph and a minimum common hypergraph according to an HCCG graph of the extracted event; 3) performing incremental clustering on HCCG through context information of social media, and gradually highlighting event elements of news in a clustering process; and 4) performing event judgment through an HCCG model-based clustering result, and judging whether the clustering result is a true event or not. According to the method, dispersed social media information can be effectively collected; intermediate and final event detection results are expressed in a multi-granularity manner visually by using an entity relationship model; and compared with a conventional social media event extraction method, the social media graph representation model-based social risk event extraction method has better generalization application capability and higher accuracy.

The invention discloses a multi-modal feature selection and classification method based on a hypergraph. The hypergraph can be used for effectively carrying out modeling on the high-order information of data. In the method disclosed by the invention, firstly, the hypergraph is independently constructed for the sample of each modal, and the Lapras matrix of the hypergraph is calculated; then, each group of modals is taken as one group of tasks, and a 12,1 norm is used for carrying out feature selection to guarantee that the characteristics of the same cerebral region of different modals can be selected. In addition, a hypergraph regularization term is used for describing high-order information between data samples so as to fully utilize a distribution prior in each group of modal data. Finally, a multi-core support vector machine is used for carrying out fusion classification on selected characteristics. By use of the method, the feature selection and classification can be effectively carried out.

The invention provides a video semantic analysis method based on a self-adaption probability hypergraph and incremental semi-supervised learning. The video semantic analysis method based on the self-adaption probability hypergraph and the semi-supervised learning comprises the steps that (S1) a hypergraph model is established by means of a self-adaption probability hypergraph establishment method, (S2) the semi-supervised learning is conducted on the hypergraph model by means of the spectrogram segmenting principle, (S3) a semi-supervised model based on the self-adaption probability hypergraph is perfected by means of an increment mechanism, and (S4) semantic analysis is conducted on a tested video by means of the perfected hypergraph model. According to the video semantic analysis method based on the self-adaption probability hypergraph and the semi-supervised learning, the establishment of the self-adaption probability hypergraph and an incremental semi-supervised learning method are combined for use, the sensibility to a radium parameter when an ordinary hypergraph model is established is eliminated, and the accuracy and the robustness of the model are improved; in addition, under the incremental semi-supervised learning mechanism, semantic searching accuracy and semantic searching completeness are improved remarkably.