176 results about "Graph match" patented technology

Example embodiments of systems and methods for network pattern matching provide the ability to match hidden networks from noisy data sources using probabilistic multi-attribute graph matching analysis. The algorithms may map roles and patterns to observed entities. The outcome is a set of plausible network models. The pattern-matching methodology of these systems and methods may enable the solution of three challenges associated with social network analysis, namely network size and complexity, uncertain and incomplete data, and dynamic network structure.

An information searching apparatus retrieves a sub graph matching an inquiry graph from a graph to be searched. The apparatus includes an extracting unit that extracts, from among clusters of nodes in the graph to be searched, plural cluster pairs that each include a first cluster and a second cluster including a node linked by a link to a node in the first cluster and a calculating unit that calculates a bonding strength for each of the cluster pairs. The apparatus further includes a determining unit that determines, among the cluster pairs and based on the bonding strength of each of the cluster pairs, a cluster pair to be merged; a merging unit that merges the cluster pair; and a searching unit that searches the merged clusters for a sub graph matching the inquiry graph. An output unit outputs a search result of the searching unit.

The invention discloses a RFID (radio frequency identification) technology based learning machine for the early education and interaction of children and a learning method thereof. A RFID reader is respectively connected with a recognition processing module and an antenna, the recognition processing module is respectively connected with a voice broadcast module, a voice recognition module, an SD card read module and a battery module, a display module is also respectively connected with the recognition processing module and the battery module, and a RFID tag card is communicated with the RFID reader through a radio frequency signal. In learning and entertainment modes, dynamic graphs matched with pictures on the surface of the RFID card are circularly displayed on a screen of the learning machine; the RFID reader reads tag information in the card, transmits obtained recognition codes to the recognition processing module to recognize, and describes the information of the pictures on the surface of the RFID card in video playing cartoon image and voice broadcast forms; and in a recognition test mode, pictures of training aids are displayed on the screen, then the names are talked about with the given pictures, the recognition processing module carries out signal processing and compares the name information of the pictures on the screen, and then results are displayed. The learning machine has the characteristics of simple structure, easy operation and high feasibility.

The invention discloses a remote sensing image classification method based on active deep learning. The remote sensing image classification method includes the steps that (1), remote sensing image data to be classified are selected; (2), the remote sensing image data are processed by utilizing an algorithm configured in advance; (3), the optimal sample B is selected from unmarked samples U by applying the active learning algorithm nEQB; (4), the optimal sample B is subtracted from the unmarked samples U to obtain a new unmarked sample set U', and the optimal sample B is added to marked samples L to obtain a new marked sample set L'; (5), the step (2) is executed again, the process continues to be circulated, the circulation is completed until the unmarked sample set U'is a null set or a preset learning stopping index is met, and classification accuracy and a classification result graph matched with the classification accuracy are output. The remote sensing image classification method has the advantages that through deep learning and active learning, the defects caused by using unsupervised learning and supervised learning can be overcome, and the classification accuracy of the data is effectively improved.

The invention discloses an industrial graph fusion method based on a graph convolutional neural network. The method is based on a plurality of constructed industrial sub-graphs, and the method comprises the following steps: constructing a local entity sub-graph of a graph; converting the structure embedding of the entity into the same vector space by using the attribute embedding of the attributetriple in the graph, forming an entity embedding vector, converting an entity alignment problem into a graph matching problem, and further forming a local matching vector by using a graph attention method; propagating local matching information in a graph through GCN to form a graph-level matching vector, and finally obtaining entity alignment in the graph through a double-layer feedforward neuralnetwork. According to the invention, the structural embedding of entities is converted into the same vector space through attribute embedding, the problem that pre-aligned entities are insufficient is solved, and the entity alignment problem in the graph is further converted into the graph matching problem through graph attention. Intelligence support is provided for optimizing the industrial structure, optimizing the regional structure and improving the industrial core competitiveness.

The invention provides a software definition network message processing method. The software definition network message processing method is characterized in that a forwarding flow graph request, which can be reported because of inability of matching a data message forwarded by a host device according to requirements with a forwarding flow graph, can be received; when the target MAC address of the data message is the global virtual MAC address, the forwarding flow graph can be generated according to the global port; the forwarding flow graph comprises the operation instruction of replacing the target MAC address of the data message by the MAC address of the global port graph matched with the target IP address of the data message; the forwarding flow graph can be transmitted to the host device, which can be used to forward the data message matched with the forwarding flow graph. Based on the same conception, the invention also provides a software definition network message processing device, and therefore the distributed routing can be realized, and the VXLAN gateway is not the bottleneck of the cross-network segment forwarding any more.

The embodiment of the invention discloses a screen light supplementing method and device, a storage medium and an intelligent terminal. The method comprises the steps of when it is monitored that a screen light supplementing event is triggered, determining a screen light supplementing mode according to the state information of a flash lamp; obtaining the parameter information of a light supplementing graph matched with the screen light supplementing mode, generating a light supplementing graph layer based on the parameter information, wherein the light supplementing graph layer comprises an opaque light supplementing graph area and a transparent preview area; and obtaining a preview interface, carrying out superposition processing on the light supplementing layer and the preview interface,and displaying the superposed preview interface so as to display the preview image through the preview area when the screen light supplementing is carried out. According to the scheme, different light supplementing graphs can be displayed according to different screen light supplementing modes, and the problem that in the prior art, the light supplementing requirements of different scenes cannotbe met due to the fact that the fixed pure color images are adopted for light supplementing is solved.

The invention provides an embedded terminal software code vulnerability detection method and device based on a neural network, and the method comprises the steps: obtaining a source code of target embedded terminal software, and carrying out the preprocessing of the source code to obtain a binary code; inputting the feature function of the binary code into a pre-trained neural network to obtain anattribute control flow graph of the binary code; inputting the attribute control flow diagram of the binary code and the attribute control flow diagram of the vulnerability code into a Siamese network for similarity comparison to obtain a comparison result; wherein the vulnerability code is a known vulnerability in a preset vulnerability library; and finally, determining whether the source code of the embedded terminal software has a known vulnerability or not based on a comparison result. According to the method, the universality of detection is improved by converting the source code into the binary code through preprocessing, meanwhile, the limitation of a graph matching algorithm is overcome by generating the attribute control flow graph based on the neural network, and the detection efficiency is improved.

The invention provides a method for automatically locating and moving a binocular vision mechanical arm for weight testing. A camera is calibrated, and a feature template is determined; the camera collects an image of a weighing apparatus centring disc, an image of a weight to be tested and an image of a mechanical arm tail end clamp, and feature graphs matched with the feature template are found from the obtained images; the positions of the feature graphs under a pixel coordinate system are calculated and converted into positions under a fixed coordinate system; a first path is generated according to the positions of the tail end clamp and the weight to be tested, and the mechanical arm completes the work of clamping the weight to be tested according to the first path; a second path is generated according to the positions of the weight to be tested and the weighing apparatus centring disc, and the mechanical arm moves the weight to be tested to the target point of the weighing apparatus centring disc according to the second path. Locating precision can be improved, the carrying efficiency of the weight can be improved, labor cost can be lowered, and impact damage to a weighing apparatus is avoided.

The invention provides a case query method based on a knowledge graph, a case query device, computer equipment and a computer readable storage medium, and the case query method comprises the steps: determining case information corresponding to a query instruction when receiving the query instruction for an appointed case; determining a case type of the case information according to a preset classification model so as to determine a pre-stored case map set corresponding to the case type in the map database; executing map extraction operation on the case information to generate a to-be-matched knowledge map; according to a preset graph matching algorithm, detecting whether a pre-stored case graph set has a pre-stored graph matched with the to-be-matched knowledge graph or not, and when it isdetected that the pre-stored graph exists, determining corresponding similar case information according to the pre-stored graph. Through the technical scheme of the invention, the time for determining the similar case information is effectively reduced, and the accuracy of the determined similar case information is improved.

The invention discloses a PPI network comparison-oriented graph matching constraint solving symbol method. According to the method, a CSP model is established (a constraint condition with a shortest path is established) by utilizing a constraint relationship in a graph structure; by adoption of a graph-based backtracking algorithm, an OBDD symbol technology and each included symbol operation, theaim of solving subgraph isomorphism problems is achieved; and finally, the technology is imported in to a PPI network comparison problem, the problem is solved and a PPI network comparison-oriented graph matching constraint solving symbol technology is given. According to the method, the graph-based backtracking algorithm which is used for solving constraint satisfaction problems is combined, theOBDD symbol technology is adopted, advantages of an operation method are developed, and a graph matching symbol algorithm is applied to the protein mutual action network comparison problems in the field of biological information according to a constraint symbol solving technology which is used for solving sub-graph isomorphism, so that the problem solving efficiency can be improved to a certain extent and the state space complexity is reduced.

The invention provides a graph drawing method and device which are applied to terminal equipment, and the terminal equipment comprises canvas deployed on the upper layer of a video stream display window and structured feature data of at least one basic graph matched with the canvas. The method comprises the steps of monitoring whether a target graph drawing event performed by a painting brush based on a selected basic graph is generated or not in real time; when a target graph drawing event is monitored, obtaining drawing coordinates of a drawing pen at the current time point at the time pointwhen each frame of image in the video stream is displayed respectively; and when the drawing coordinates are obtained, rendering and drawing a target graph corresponding to the image displayed at thecurrent time point on the canvas according to the drawing coordinates and the structured feature data of the selected basic graph. The method is small in data processing amount and high in data processing efficiency, graph drawing can be carried out on the video played and displayed by the browser, the graph drawing animation effect is achieved, and it is ensured that the drawn graph can be interactively processed.

The invention provides a floor plane graph generation method and device, and relates to the technical field of home decoration construction drawings. The method comprises the following steps: determining wall body information according to house type design data, wherein the wall body information comprises wall body thickness and two endpoint coordinates of a wall body center line; determining a wall body center line connection relation according to the coordinates of the two end points of the wall body center line; scaling the first wall body center line according to the wall body thickness toobtain a surface line segment of the first wall body, the first wall body center line being any one wall body center line included in the house type design data; and according to the wall body centerline connection relation and the surface line segment of the first wall body, adjusting the redundant line segment at the joint to obtain a floor plane graph matched with the floor plane design data,the wall body joint being the surface line segment where the first wall body and the second wall body intersect, and the second wall body being the wall body having the connection relation with the first wall body. According to the method, the floor plane graph used for construction is automatically generated based on the house type design data, and the drawing efficiency and accuracy of the construction floor plane graph are improved.

The invention provides a road scene matching method and device. The method comprises the steps of inputting a road scene picture into a target detection module, and obtaining a target detection resultcorresponding to the road scene picture output by the target detection module; based on a deunay triangulation algorithm, generating an adjacency relation graph corresponding to the road scene picture according to a target detection result; inputting the adjacency relation graphs corresponding to any two road scene pictures into a graph matching module, and obtaining the similarity between the two road scene pictures output by the graph matching module; and if the similarity is greater than a preset threshold, performing feature matching on corresponding targets in the two road scene picturesto obtain a matching result. According to the method and device provided by the embodiment of the invention, the scene matching precision is improved, the speed is far higher than that of a traditional scene matching method, manual setting of selection parameters is not needed, the method and device can be widely applied to road scenes, and meanwhile, the relevance problem of a deep neural network is solved.

The invention discloses a remote sensing image change detection method under low registration precision based on a graph matching model. The method comprises the following steps: firstly, carrying outbuilding segmentation by using a multi-scale adaptive convolution remote sensing image semantic segmentation model to obtain segmentation results of different time phases in the same region, namely binary images; in order to solve the problem of region connection after segmentation of a plurality of different buildings, adopting plaque extraction based on an instance segmentation method, so thatdifferent plaques are distinguished, and regions corresponding to the plurality of buildings in different time phases are obtained; carrying out patch matching based on a graph matching model according to the position information corresponding to the patch and other characteristics; and finally, extracting corresponding change information according to the matched information. According to the invention, high-precision change detection of remote sensing images with different time phases under low registration precision is realized. The change detection scheme provided by the invention does notdepend on accurate image preprocessing, especially image correction and registration, and the requirements for image correction and registration are reduced.

The invention discloses a cardiac motion representation method and device and a computer readable storage medium. Extracting the MR image of the first phase and the second phase to be processed and constructed so as to obtain a first phase cardiogram structure A and a second phase cardiogram structure Bi; a vertex set with the corresponding relation is obtained by solving the graph matching convexobjective function by using the two cardiogram structures; a deformation function is constructed based on the vertex set with the corresponding relation, and deformation of Bi is performed by using the deformation function so as to obtain a new second phase cardiogram structure Bi+1; and calculating the change value of Bi and Bi+1 and judging whether the change value is greater than the preset threshold value, if so, i=i+1 and circulating the steps, or using the deformation function to represent the cardiac motion. According to the method, the deformation function is constructed by the cycling mode and then the original structure is deformed by using the constructed deformation function, and the accuracy of the deformation function is further checked according to the deformation change value so that the method can more accurately represent the cardiac motion.

The invention discloses a node-graph cross-layer graph matching system based on a graph neural network. The system comprises a graph node embedding vector encoding module, a node-graph cross-layer graph matching network module, a graph vector encoding module and a similarity calculation module. In addition, the invention further discloses a node-graph cross-layer graph matching method based on thegraph neural network. The method is based on a graph neural network. Learning cross-level interaction information of each node in one graph and the whole other graph in a fine-grained manner; according to the method, the embedded information of each node can be effectively updated, and a cross-layer graph matching system based on the method can be obtained by training based on an end-to-end mode,so that the similarity between any two graph data can be quickly and accurately calculated.

The invention discloses a graph matching based cross-species biological access discovery method, which can solve the problem that the conventional biochemistry experiment methods are low in biologicalaccess discovery efficiency, and the problem that the conventional graph matching methods cannot integrate the biological sequence similarity and the protein interaction network structure similaritytogether. In the method, the biological sequence similarity and the protein interaction network structure similarity can be integrated together, coexisting larger sub structures in protein interactionnetworks of different species can be discovered, biological accesses having the similar functions in the different species can be effectively discovered, and the method has guiding significance for research on relation between different species in biology.

The invention discloses a sub-graph matching device, which is used for finding out a sub-graph matched with a loop graph in a large-scale graph. The device comprises a spanning tree module, a matching module, a judging module and a gathering module. A spanning tree of the loop graph is found by the spanning tree module according to a minimum spanning tree algorithm; the spanning tree matching is carried out on a data graph by the matching module from bottom to top to find a tree Li (i>=1) matched with the spanning tree; whether the sum of a grading function value of the tree Li and the missing edge number is larger than or equal to a preset value or not is judged by the judging module, if so, whether the tree Li can be expanded into the loop graph or not is judged by the judging module; and the gathering module can be used for outputting the previous fixed value number of matching graphs in a set V when the sum of the grading function value of the tree Li and the missing edge number is larger than or equal to the preset value.

The invention proposes a pedestrian re-identification method and system based on key point optimization and multi-hop attention graph convolution, and the method comprises the steps: extracting key points in an image, obtaining a confidence level corresponding to each key point, and obtaining a pedestrian re-identification result based on the low confidence level of the corresponding key point when a pedestrian in the image is shielded; approximately estimating feature information of the shielded key points by using corresponding features of other key points for the key points with low confidence levels; for the processed feature information of the key points, using an added feature learning network with a multi-hop attention mechanism for learning, considering effective information of each layer in the network, combining with an attention matrix with the feature information to acquire output of feature learning; and for the output features, enabling the feature matching network to compare the node relationship in the two images through a graph matching algorithm and the human body topological features, and output a predicted classification result. According to the invention, the precision of re-identification of shielded pedestrians is improved.