82 results about "Graph mapping" patented technology

A method of analyzing structure of a network of vessels in a medical image, comprising:

• • receiving the medical image depicting the network of vessels;
  • obtaining a mask of the network of vessels in the image; and
  • generating a non-forest graph mapping a plurality of paths of vessels in the network to directed paths in the graph, with each edge in the graph either directed to indicate a known direction of flow in the corresponding vessel, or undirected to indicate a lack of knowledge of direction of flow in the corresponding vessel, and with all directed edges in a path directed in a same direction as the path.

The invention discloses an automatic service generating system and an automatic service generating method both of which are based on an imaging service scene and a field template. The automatic service generating system comprises a browser, a server and a database. When customizing software or a service, the user drags a graph tag in a graphic interface of the browser firstly and then establishes an application scenario of the software or the service. The system analyzes the application scenario into two description files including a service procedure and a user demand through a graph mapping library, finds the field template meeting the requirement of the user in a field template library through the analyzing result and carries out dynamic customization or extension on the field template. Through searching and testing, components which are found in a component library and can achieve the setting purpose are added into the field template which is customized and includes the software or the service procedure structure and the service description information of the service procedure structure, and the process of generating the software/the service is achieved. At last, the testing and deployment are carried on the generated software/the service to enable the user to be accessed to or to use the customized software or the service. The automatic service generating system is easy and convenient to operate and maintain, and the user coverage is wide.

A method for service creation based on graph mapping including constructing a graph having leaf nodes in an environment including business support nodes associated with respective services and operations support nodes including attribute models, identifying a plurality of services at a level of the leafs in the graph, extending the graph by representing existing services offerings in terms of operation level service definitions and new services at the level of the leafs, and re-combining leaf nodes to define new services at service creation time.

The invention relates to a text classification method based on a graph kernel and a convolutional neural network, and belongs to the technical field of data mining and information retrieval. The coreconcept is that: firstly, a text is preprocessed into an expression mode of a graph structure, wherein nodes in a graph correspond to words in the text; then weights of the nodes are calculated on thebasis of the graph structure, then the graph structure is decomposed into a plurality of sub-graphs by using a community discovery algorithm, the graph is mapped to a high-dimensional space by usinga graph kernel technology so as to obtain a tensor expression of the graph; finally, the tensor expression is input into the convolutional neural network, deep mining is carried out on graph features,and a category of the text is output. Compared with the prior art, according to the text classification method, an internal structure and context semantics of the text can be sufficiently utilized, so that text contents are sufficiently expressed; node information is more reasonable; the problem of complex and fussy processing process in the text classification is effectively solved.

The invention discloses a block adaptive type Cartesian grid rapid graph mapping method and system, and the method comprises the following steps: S1, after an MPI environment is initialized, importing geometric data of a distributed block adaptive type Cartesian grid into each processor so as to reconstruct a multi-way tree structure, and building a link relation of adjacent hierarchical nodes; S2, sharing node geometric information at a region decomposition boundary across processors, and storing the node geometric information in a corresponding virtual node data container; S3, constructing a local geometric topology of each leaf node on the current processor; S4, normalizing the completely overlapped geometric elements on each processor; S5, based on cross-processor query operation, establishing dynamic communication mapping of docking surface elements at the region decomposition boundary; and S6, exporting the non-structural polyhedron computational grids in parallel on each processor. The problems that in the prior art, a self-adaptive geometric module and a physical field numerical simulator are difficult to fuse, and the dynamic geometric data format conversion efficiency is low are solved.

The invention discloses a fault accurate identification method based on real-time intra-station graph mapping. A fault vector analysis and fault diagnosis platform based on a primary topological graph scalable vector graph and a universal information model is built; fault equipment substations needing fusing are determined for a search route through a fault amount flow direction, and a macrozone information fusion platform is formed; information fusion is carried out based on the macrozone information fusion platform, and an information closed loop and an oscillograph are fused into a vector analysis graph to carry out accident property analysis and assistant decision making; in the process of accident property analysis and assistant decision making, fault information is collected and fused in an expanded macrozone, and the fault property of a fault element is judged. On a fault scene, intra-station equipment serves as the background, the flowing direction of fault current serves as an analysis interface of a display mode, judgment of the fault element can be achieved rapidly, and the fault judgment on a multi-element connection device with a converge function is especially visual and effective.

The invention discloses a degree distribution histogram publishing method capable of meeting node differential privacy. The method comprises the step of carrying out graph mapping by adopting an edgeadding sequence; based on the observation, carrying out priority ranking based on the degrees of the nodes corresponding to the edges, wherein the smaller the sum of the degrees of the nodes is, the higher the priority is. The graph mapping optimization method disclosed by the invention has the beneficial effects that more edge information can be reserved, and the change of node degree is reduced.Meanwhile, a new histogram bucket division strategy is provided to reduce the influence of random noise added in the histogram publishing process on original data distribution.

The invention belongs to the field of integrated circuits and relates to a graph isomorphism judging method based on spectrum analysis. The method includes: large-scale pure-resistor network graphs are modeled into non-mixing undirected simple graphs, two-dimensional planar graphs are mapped into one-dimensional distribution, and isomorphism of two graphs is judged according to processed one-dimensional distribution conditions. The method has the advantages that the method is accurate and fast in judging of the non-mixing undirected simple graphs, especially the large-scale non-mixing undirected simple graphs; the method is evidently faster than a Nauty method which is good in performance currently, and the method is well applicable to fields such as judging of identical sub-circuits in large-scale integrated circuits and judging the isomeride in organic chemistry.

The invention belongs to the technical field of neural network models, and particularly relates to a neural network weight initialization method based on transfer learning. According to the method, for a specified target task, a neural network model, namely a teacher model, with high complexity is designed, the teacher model is trained, and after training is completed, a generated feature graph isused for guiding weight initialization of a student model; the difference between the feature maps is calculated or the feature maps are mapped into a regeneration kernel Hilbert space, the difference of the feature maps in the regeneration kernel Hilbert space is calculated, and the calculation is simplified by adopting a kernel function method. According to the method, a simple student model isenabled to achieve a better weight initialization effect, and general training is performed on the student model after weight initialization is completed so that the student model is enabled to achieve a better global convergence point and the performance of the student model is enabled to be more excellent. According to the method, the performance of the student model can be effectively improvedon the premise of not increasing the complexity of the student model.

The invention discloses an unconstrained face recognition method based on weighted tensor sparse graph mapping, and relates to the technical field of face recognition methods. In the sparse graph construction stage, training samples (images) are represented by second-order tensors, a supervised over-complete tensor dictionary is constructed, and similar sparse reconstruction coefficients of the samples are optimized and solved; and a more accurate tensor sparse neighbor graph is constructed in a self-adaptive manner. In a bilateral low-dimensional projection stage, low-dimensional tensor subspace distribution is obtained by utilizing identification information implied in sample global distribution. And low-dimensional mapping yWTSGE = UTyV is performed on the to-be-tested sample y by adopting the optimal WTSGE bilateral projection matrixes U and V, and a classifier is trained by using a low-dimensional training sample DWTSGE = UTXV to realize accurate identity authentication of the non-constrained face. According to the method, the complexity of the non-constrained face image data is fully considered, the neighbor distribution diagram of the high-dimensional tensor data is obtainedin a self-adaptive mode through the sparse representation technology, the low-dimensional manifold essential structure of the highly-distorted non-constrained face data is effectively extracted, andthe accuracy of non-constrained face recognition is greatly improved.

The invention provides a cylinder sleeve defect mark detection method based on deep learning, and the method comprises the following steps: collecting a cylinder sleeve image to construct an original data set, carrying out the preprocessing and marking of the images in the original data set, and obtaining a local defect graph of a training set; carrying out modeling by using a Mask R-CNN algorithm based on a Swin converter to obtain a network model; detecting the images of the test set through the obtained network model; extracting a region of interest from a detection result through a mask mechanism to enhance a detection effect; evaluating the detection performance of the network model; and performing defect mark detection on the acquired cylinder sleeve image through the obtained network model. According to the invention, through image preprocessing, cylinder liner noise influence reduction, local graph construction of an original graph, obtaining of a new batch of training data, through a method of mapping a large graph with a small graph, obtaining of a detection result of the original graph, through a mask mechanism, filtering of a noise area, and improvement of small target detection precision.

The invention discloses a method for detecting a local long and narrow land utilization vector graph. The method comprises the following steps of: negatively buffering an original graph; calculating atheoretical difference value of the perimeter of the graph before and after buffering; calculating an actual difference value of the perimeter of the graph before and after buffering; and judging thelocal long and narrow pattern spots according to the relationship between the perimeter theoretical difference value and the actual difference value. By means of the method, on one hand, a formula for effectively calculating the theoretical difference value of the perimeter of the graph before and after buffering is provided, and the view angle and method of graph research are expanded; and on the other hand, the local long and narrow graph can be automatically detected through a computer program without depending on manual visual identification, so that non-standard graph mapping is effectively avoided, and the data quality is improved.

The invention discloses an application system performance bottleneck determination method, and relates to the technical field of computer networks, and the method comprises the steps: connecting a server, network equipment and storage equipment based on a preset network connection model, so as to generate a physical topological graph; mapping the physical topological graph into a service topological graph based on a service mapping model; and obtaining target performance data indexes of the server, the network equipment and the storage equipment according to the physical topological graph and the service topological graph, analyzing log information corresponding to the target performance data indexes, and determining the performance bottleneck of the application system according to the log information. According to the invention, the performance bottleneck of the application system is determined through the monitored log information of all performance fault suspicious points, and analysis is carried out according to the performance bottleneck of the server, the network equipment and the storage equipment, so that the performance analysis time is greatly shortened, the fault positioning and solving efficiency is improved, and the user experience is improved. And the maintainability and the stability of the application system are further improved.

The invention discloses a pressure gauge calibration table reliability evaluation method based on a Bayesian correction model. The method comprises the following steps: establishing a fault characteristic index system capable of representing a pressure gauge calibration table; constructing an accident tree model of the pressure gauge calibration console; establishing a BN model graph of the pressure gauge calibration console, and directly mapping the FTA model graph into the BN model graph through a graph mapping method; determining a prior probability of a bottom event and a conditional probability between layers; correcting the Bayesian model according to the actual situation; calculating a failure probability; and calculating the posterior probability of the bottom event, and searchingthe maximum risk bottom event causing failure. Compared with a traditional simplex qualitative and simplex quantitative reliability evaluation method, the method has the advantages that the accuracy is high; fuzzy mathematics and quantitative data analysis are fused, the defects that qualitative analysis lacks data support and pure data analysis is too absolute are overcome, and compared with a traditional Bayesian network model, the variable correction process of the Bayesian network is added, and the obtained evaluation result can be closer to the actual engineering situation.

A graph width learning classification method based on global sampling sub-graphs comprises the following steps: 1) global sampling: performing global sampling on an original network by using a connected edge sampling method to obtain sub-graphs; 2) sub-graph mapping: mapping the sub-graphs once or twice through a mapping mechanism from connecting edges to nodes to obtain a first-order network and a second-order network; 3) graph feature extraction and fusion: extracting features of an original network and all mapped networks through a Graph2vec model, and then splicing the extracted original network features and first-order and second-order network features after each sampling mapping as feature representation of an original graph; and 4) classifying a width network. The invention further discloses an efficient and accurate graph classification system based on the method. A weight matrix in a width network classifier is trained in a supervised manner by combining the fused graph features and the known graph labels. Finally, effective classification of the graphs is realized according to the weight matrix in the width network and the input graph features.