24 results about "Big graph" patented technology

The invention discloses a causal evaluation method for advertisement putting. The method comprises the following steps: acquiring a plurality of data sets of advertisements put on a take-out platform by merchants operating on the take-out platform; inputting the plurality of data sets into a causal graph mining model set to obtain a first candidate causal graph set; filtering the first candidate causal graph set based on an expert knowledge base to obtain a second candidate causal graph set; obtaining the weight of the path according to the relationship between the node and the path of each second candidate causal graph in the second candidate causal graph set, and aggregating the plurality of second candidate causal graphs based on the weight of the path to obtain a causal assessment large graph; and based on the causal evaluation large graph, evaluating the causal relationship between the advertising parameters and the effect indexes of the merchant advertising. According to the method, the accuracy of advertisement effect evaluation can be improved through accurate causal relationship analysis, the advertisement putting efficiency can be improved, and the putting strategy can be optimized.

A method is disclosed. The method includes generating a graph with nodes and edges, inputting the node data for the nodes into a first scoring algorithm to obtain node scores, and inputting the edge data into a second scoring algorithm to obtain edge scores. The method also includes forming one or more subgraphs from the graph. Each subgraph is formed by: determining an initial node with a node score that exceeds a threshold, determining an initial set of edges connected to the initial node, determining a set of nodes connected to the initial set of edges, and determining additional edges and additional nodes extending from the set of nodes, until a predetermined number of the additional nodes and edges have edge scores and node scores that do not meet the threshold, thereby forming the subgraph.

The invention discloses an improved mosaic data enhancement method, computer equipment and a storage medium. The method comprises the following steps: creating a background large image, and splicing a plurality of original images containing a target frame at the central point of the background large image to obtain a spliced large image; creating a window with the same size as the training image, and sliding on the spliced large image to obtain a plurality of candidate areas; calculating the score of each candidate region based on the score weight corresponding to the category and the area of each target frame in each candidate region; and performing soft sampling on the plurality of candidate areas with the highest scores, and cutting the optimal candidate area to generate a training image. Through the above mode, the category of the target frame can be considered as the score weight, the sampling probability of the tail category is improved, and the recognition performance in a long-tail scene is improved; and the area of the target frame is considered as a score weight, the integrity rate and the size of the target frame in the candidate region are evaluated, the contribution of an incomplete sample is inhibited, and the quality and the robustness of a training image are improved.

The invention discloses a data warehouse system based on a graph database and a construction method, the system comprises a graph data storage layer and a platform service layer, the graph data storage layer comprises a physical large graph storage module and a business sub-graph storage module; the physical large graph storage module adopts a distributed graph storage engine and is responsible for storing enterprise core data elements and association relationships thereof; a graph database is adopted as a bottom storage facility; the service sub-graph storage module provides an independent sub-graph storage space for each service application based on the data of the physical large graph storage module; dynamic graph mode expansion is supported, business departments are allowed to add specific attributes and relationships, and customized business sub-graphs are formed; the platform service layer extracts data from each business system through a high-performance ETL engine provided by a graph database, and stores the data to the physical large graph storage module after data governance service cleaning and conversion; the method has the beneficial effect that the flexibility and adaptability of business application are remarkably improved.

The application relates to the technical field of engineering simulation, and discloses an industrial structure mechanics simulation prediction system based on an enhanced graph attention network, which comprises a finite element data mapping and enhancement module, a prediction module, an optimization module and a deployment module. The system reconstructs a finite element grid into a graph structure tensor, captures geometric mutation characteristics by using high-dimensional feature projection and adaptive feature recalibration, and transmits deep physical information through dense residual connection. In the training stage, a physical constraint loss function containing attention area weighting and structure smoothness is introduced; in the reasoning stage, a mixed precision mode is adaptively switched according to hardware attributes, and large graph blocking and fusion calculation based on overlapping boundaries are performed. The application effectively solves the contradiction between calculation efficiency and numerical accuracy in complex industrial structure mechanics response prediction, and realizes millisecond-level high-fidelity online simulation.

According to the phishing account detection model training method, the phishing account detection method and the phishing account detection device, a second-order transaction network is independently constructed for each account, a local transaction structure of each account is completely reserved, and the bottleneck of feature dilution and excessive smoothness caused by traditional global large graph training is broken through; address similarity features are introduced, high-frequency transaction object tail number similarity deliberately constructed by an attacker is accurately captured, and the visual confusion type fishing behavior is effectively recognized. A random walk restart algorithm is adopted to generate double local subgraphs, center node self-pairing is used as a positive sample, a cross-subgraph non-center node is used as a negative sample, node-level comparison loss is constructed, and the discrimination capability of a model on phishing nodes and normal neighbors is enhanced; through joint classification loss end-to-end training, the graph neural network can be incrementally updated without re-training a global graph, and lightweight deployment is realized. According to the method, in a block chain transaction scene with extremely unbalanced data, the risk of misjudgment of the phishing account can be remarkably reduced, and high sensitivity and real-time performance are both achieved.

1. Name of the product in this design: Graphical User Interface for Viewing Activity Calendar on Electronic Devices. 2. Purpose of this design: An electronic device. 3. The key design features of this product are its graphical user interface content. 4. The image or photo that best illustrates the design points: Interface change state diagram 1. 5. The rear view, top view, bottom view, left view, and right view are standard design elements and are omitted. 6. Purpose of the graphical user interface: The graphical user interface is used to view and subscribe to activities at different times in a calendar format. In the main view, click the bubble area under the date in the calendar widget to enter the interface change state shown in Figure 1. The top displays the activity calendar for the current month, and the bottom displays the activity content. In the interface change state diagram 1, swipe up on the activity content area to enter the interface change state diagram 2 and switch to the weekly calendar format. Scroll through the activity content area in Interface Change State Diagram 2, and sequentially enter Interface Change State Diagram 3 and Interface Change State Diagram 4 to display activities for future dates. The icons in the calendar area switch left and right according to the activity dates. In the interface change state diagram 4, scroll through the activity content area to enter the interface change state diagram 5, which displays future subscription coupons and other activities. The content in the calendar area switches according to the week of the date. In the interface change state diagram 5, continue scrolling through the activity content area to sequentially enter the interface change state diagrams 6, 7, and 8, displaying activity content with different numbers of images. In the interface change state diagram 8, click the "Remind Me" button in the lower right corner to subscribe to the event, and then enter the interface change state diagram 9. In the interface change state diagram 9, clicking on any image in the activity will lead to the interface change state diagram 10, where the image is displayed in a large format. Users can zoom in, zoom out, switch between images left and right by swiping the screen, or save the image by long-pressing.

The application relates to a dense subgraph-based graph compression method and system, and belongs to the technical field of computer software. The application reorders and reencodes nodes by using dense subgraphs, so that node ordering is not performed on a large graph, time for node ordering is reduced, and redundancy consumption of each node in the memory is reduced; each node in a K-edge-connected subgraph is re-encoded, a new edge storage representation is constructed, and space occupied by nodes with high occurrence frequencies in the graph is reduced; for a sparse space of the graph, a compressed adjacency list is used to reduce storage consumption, and a two-level index is used to improve the search efficiency of node neighbors. The application is simple and easy to use, reduces the gap distance between nodes, and effectively reduces the space required for storing the graph in the memory.

In order to solve the problems that in the prior art, a social network and a knowledge graph have inherent contradictions in the aspects of dynamics, heterogeneity, time sequence and multi-source data fusion, and real-time modeling and analysis of risk knowledge flow in a financial scene cannot be met, the invention discloses a social knowledge large graph construction method facing financial risk knowledge flow. According to the method, multi-source data such as financial markets, financial news and financial platform user social relations can be collected, point sets are extracted, edge sets and three-party hyperedge sets related to enterprises, social users and financial news are constructed, and after an initial social knowledge large graph is formed through integration, a graph structure is dynamically updated based on time range changes; according to the method, deep fusion of social relations and financial field knowledge is realized, propagation paths and intensity changes of financial risk knowledge along with time are accurately captured, the problem of multi-entity association analysis sparsity of a traditional model is solved, and financial risk monitoring and early warning requirements are met.

This invention provides a method and system for subgraph matching in large graphs based on multiple GPUs. The method includes dividing a large graph into subgraphs to obtain several subgraphs, including intra-partition subgraphs and cross-partition subgraphs; performing sequential queries on the intra-partition subgraphs to obtain a matching order, and determining a first matching strategy for the intra-partition subgraphs based on the matching order; identifying the hop count of the cross-partition subgraphs and determining a second matching strategy for the cross-partition subgraphs based on the hop count; and integrating the first and second matching strategies through GPUs to complete the matching of the intra-partition subgraphs and the cross-partition subgraphs. This invention can process the subgraph matching problem in parallel on multiple GPUs, overcoming the limitation of graph size on GPU subgraph matching in large graphs.

The invention provides a perspective line accurate concentric plotter, which relates to the field of plotters and is characterized by comprising an L-shaped ruler body, a sliding ruler and a knob. The method has the advantages that the design principle is visual, operation is convenient and fast, a complex small graph amplification process is not needed, dependence on a limited fixed template is also eliminated, and drawing can be directly conducted on a target drawing; through the precise arc track and the sliding ruler structure pointing to the circle center, it is ensured that all perspective lines can be precisely intersected at the same point, and high-precision drawing is achieved; meanwhile, the device is high in flexibility, is not restrained by a preset angle, can adapt to any visual angle and vanishing point position, and greatly improves the freedom degree and efficiency of hand drawing creation; besides, the tool is wide in application range, supports drawings of different specifications, is convenient for artists to directly draw full-size perspective large drawings and add details, and has good practicability and expansibility.

The application discloses a method, system, device and medium for building a data visualization large screen based on an improved ResNet algorithm, and the method comprises the following steps: constructing a training data set, training an improved ResNet18 model by using the training data set; removing two full connection layers of the trained improved ResNet18 model to serve as a feature extractor; extracting features of the training data set by using the feature extractor, training an SVM classifier by using the extracted features; cutting a large design graph to obtain a plurality of small graphs, inputting features of the small graphs into the trained SVM classifier after applying the feature extractor to extract the features of the small graphs, classifying the small graphs, matching components from a component library according to the categories of the small graphs, and rendering the components to a large screen according to the information of the small graphs; and configuring and adjusting the components and page attributes of the large screen according to data, display and event requirements of the large screen, so as to obtain a complete data visualization large screen. The application has a higher chart image classification accuracy than existing models, and reduces the design complexity of the large screen through low-code components, thereby avoiding a repeated configuration process.

This invention belongs to the field of privacy computing technology and provides a method for constructing a multi-departmental government knowledge graph and making collaborative decisions with intelligent agents based on asynchronous federated learning. The method includes: CLIP joint embedding feature extraction, cross-departmental collaborative updating, relational reasoning and cross-departmental entity matching, comprehensive judgment and decision-making execution path matching, construction of a domain-level large graph and exception condition library, construction of an RPA rule base, and generation of automated processes. This invention achieves semantic-level fusion of heterogeneous data through seedless entity alignment and multimodal embedding technology, breaking down data silos; improves the rationality of human-machine collaboration through a two-factor threshold judgment mechanism; and enhances the automation level of processes by automatically mapping stable business rules to RPA action sequences.

1. The name of the design product: graphical user interface for personal data saving and management of electronic equipment. 2. The use of the design product: the design product is used in the operation and display of the interface of electronic equipment. 3. The design points of the design product: the graphical user interface. 4. The picture or photo that best indicates the design points: front view. 5. The use of the graphical user interface: information display and human-computer interaction for users to save and manage personal data. 6. The human-computer interaction mode of the graphical user interface: the front view displays the data saved by the user, the "go to save" button is clicked, multiple data are saved, the interface change state diagram 1 is displayed, the "save more" button is clicked, the interface change state diagram 2 is jumped to, the first item "data name" is clicked, the interface change state diagram 3 is jumped to, the "xx system" is clicked, the interface change state diagram 4 is jumped to, when there is a result, the interface change state diagram 5 is jumped to, the "save to digital business card" is clicked, the interface change state diagram 6 is jumped to, the "click to view large image" is clicked, and the interface change state diagram 7 is jumped to.

The invention relates to a color-changing stumpage identification method based on phenological change and a multi-temporal remote sensing image, belongs to the technical field of remote sensing image processing and forestry information, and solves the defects in the prior art in the aspects of color-changing stumpage identification precision, efficiency, robustness and the like. The method comprises the following steps: classifying multi-temporal remote sensing images into different phenological period subsets according to phenological periods; constructing a standardized sample library; training a window-level classification model; firstly, large-step-length coarse screening is carried out on a target remote sensing image to be detected, and then a fine judgment process is carried out; calculating a disease probability accumulated value of each pixel point, and generating a disease mask matrix; and sequentially performing connected domain analysis and morphological operation on the disease mask matrix to obtain a final disease distribution mask matrix, and outputting the final disease distribution mask matrix. According to the method, the window-level sample is used for training and reasoning, an automatic identification process from large image discrimination to high-resolution mask generation is realized through a multi-step sliding window and a probability accumulation strategy, and efficient, accurate and automatic color-changing standing tree identification is realized.

The invention discloses a remote sensing image cloud detection method based on a DeepLabV3 semantic segmentation framework. The method comprises the following steps: (1) constructing a semantic segmentation framework which comprises a data processing module, a network construction module, a model training module, a slice reasoning module, a model export module and a large graph reasoning module; (2) obtaining slices of the original remote sensing image in a data processing module by increasing the size of a cutting frame, compressing the slices to a model training size to obtain samples, and marking the samples; (3) training a DeepLabV3 model predefined in the network building module by using the sample data in the model training module; (4) if the image slices are input, reasoning is carried out through a slice reasoning module; if a large-size remote sensing image is input, the large-size remote sensing image is converted into an engine model through the model export module, and then reasoning is carried out through the large image reasoning module. During large-scale remote sensing image reasoning, a single-machine multi-card multi-process parallel reasoning method is combined on the basis of the step (4). The remote sensing image cloud detection precision and speed can be improved at the same time.

The present application relates to a method for identifying color-changing standing trees based on phenology and multi-temporal remote sensing images, belonging to the field of remote sensing image processing and forestry information technology, and solving the deficiencies of existing technologies in terms of color-changing standing tree identification accuracy, efficiency, robustness, etc. The method comprises: classifying multi-temporal remote sensing images into different phenological subsets according to phenological periods; constructing a standardized sample library; training a window-level classification model; performing coarse screening with a large step size and then performing fine determination on the target remote sensing image to be detected; calculating the disease probability accumulation value of each pixel point to generate a disease mask matrix; and sequentially performing connected domain analysis and morphological operation on the disease mask matrix to obtain the final disease distribution mask matrix and output. The present application uses window-level samples for training and reasoning, and realizes an automatic identification process from large image discrimination to high-resolution mask generation through multi-step sliding windows and probability accumulation strategies, achieving efficient, accurate and automatic color-changing standing tree identification.

1. Name of the product in this design: Graphical User Interface (Product) for Product Display and Production of Electronic Devices. 2. Purpose of this design: An electronic device. 3. The key design feature of this product lies in the user graphical interface on the screen. 4. The picture or photo that best illustrates the key design points: Design 1 front view. 5. Design 1 is designated as the basic design. 6. Purpose of the graphical user interface: This graphical user interface is a user interaction interface for product display and product creation. 7. Human-computer interaction method of graphical user interface: Design 1 The main view is the initial interface when entering the mini program. Slide the main view of Design 1 from left to right to display the Design 1 interface change state diagram 1. After sliding, a new interface will appear, displaying the Design 1 interface change state diagram 2. Slide from left to right in the Design 1 interface change state diagram 2 to display the Design 1 interface change state diagram 3. After sliding, a new interface will appear, displaying the Design 1 interface change state diagram 4. Slide from left to right in the Design 1 interface change state diagram 4 to display the Design 1 interface change state diagram 5. After sliding, a new interface will appear, displaying the Design 1 interface change state diagram 6. Click "Product" at the bottom of the Design 1 interface change state diagram 6 to enter a new interface, displaying the interface state shown in the Design 1 interface change state diagram 7. Click the left box at the top of the Design 1 interface change state diagram 7 to enter a new interface, displaying the interface state shown in the Design 1 interface change state diagram 8. Select any option in the Design 1 interface change state diagram 8 and click "OK" to enter a new interface, displaying the Design 1 interface. The interface state is shown in Figure 9 of the Design 1 Interface Change State diagram; clicking on any product in Figure 9 will enter a new interface, which will be displayed as shown in Figure 10 of the Design 1 Interface Change State diagram; clicking the option below "Select Model" in Figure 10 of the Design 1 Interface Change State diagram will enter a new interface, which will be displayed as shown in Figure 11 of the Design 1 Interface Change State diagram; sliding the page will display Figure 12 of the Design 1 Interface Change State diagram; after sliding to a certain extent, it will display Figure 13 of the Design 1 Interface Change State diagram, where different models can be selected. After selecting a model, click "OK" in the upper right corner of Figure 13 of the Design 1 Interface Change State diagram to enter Figure 14 of the Design 1 Interface Change State diagram; clicking "Submit Order" in Figure 14 of the Design 1 Interface Change State diagram will enter a new interface, which will be displayed as shown in Figure 15 of the Design 1 Interface Change State diagram; clicking on any product in Figure 15 of the Design 1 Interface Change State diagram will enter a new interface, which will be displayed as shown in Figure 16 of the Design 1 Interface Change State diagram, where you can view a large image of the product. Design 2's main view is the initial interface for entering the mini-program. Slide the main view of Design 2 from left to right to display the Design 2 interface change state diagram 1. After sliding, a new interface will appear, displaying the Design 2 interface change state diagram 2. Slide from left to right in the Design 2 interface change state diagram 2 to display the Design 2 interface change state diagram 3. After sliding, a new interface will appear, displaying the Design 2 interface change state diagram 4. Slide from left to right in the Design 2 interface change state diagram 4 to display the Design 2 interface change state diagram 5. After sliding, a new interface will appear, displaying the Design 2 interface change state diagram 6. Click "Product" at the bottom of the Design 2 interface change state diagram 6 to enter a new interface, displaying the interface state shown in the Design 2 interface change state diagram 7. Click the left box at the top of the Design 2 interface change state diagram 7 to enter a new interface, displaying the interface state shown in the Design 2 interface change state diagram 8. Select any option in the Design 2 interface change state diagram 8 and click "OK" to enter a new interface, displaying the Design 2 interface. The interface state is shown in Figure 9 of the Design 2 interface change state diagram; click on any product in Figure 9 of the Design 2 interface change state diagram to enter a new interface, which is shown in Figure 10 of the Design 2 interface change state diagram; in Figure 10 of the Design 2 interface change state diagram, click on the option below "Select Model" to enter a new interface, which is shown in Figure 11 of the Design 2 interface change state diagram; slide the page to show Figure 12 of the Design 2 interface change state diagram; after sliding to a certain extent, it will show Figure 13 of the Design 2 interface change state diagram, where you can select different models. After selecting a model, click "OK" in the upper right corner of Figure 13 of the Design 2 interface change state diagram to enter Figure 14 of the Design 2 interface change state diagram; in Figure 14 of the Design 2 interface change state diagram, click "Submit Order" to enter a new interface, which is shown in Figure 15 of the Design 2 interface change state diagram; click on any product in Figure 15 of the Design 2 interface change state diagram to enter a new interface, which is shown in Figure 16 of the Design 2 interface change state diagram, where you can view a large image of the product. Design 3's main view is the initial interface for entering the mini-program. Slide the main view of Design 3 from left to right to display the Design 3 interface change state diagram 1. After sliding, a new interface will appear, displaying the Design 3 interface change state diagram 2. Slide from left to right in the Design 3 interface change state diagram 2 to display the Design 3 interface change state diagram 3. After sliding, a new interface will appear, displaying the Design 3 interface change state diagram 4. Slide from left to right in the Design 3 interface change state diagram 4 to display the Design 3 interface change state diagram 5. After sliding, a new interface will appear, displaying the Design 3 interface change state diagram 6. Click "Product" at the bottom of the Design 3 interface change state diagram 6 to enter a new interface, displaying the interface state shown in the Design 3 interface change state diagram 7. Click the left box at the top of the Design 3 interface change state diagram 7 to enter a new interface, displaying the interface state shown in the Design 3 interface change state diagram 8. Select any option in the Design 3 interface change state diagram 8 and click "OK" to enter a new interface, displaying the Design 3 interface. The interface state shown in Figure 9 changes; clicking on any product in Figure 9 of the Design 3 interface changes will enter a new interface, which will be displayed as shown in Figure 10 of the Design 3 interface changes; clicking the option below "Select Model" in Figure 10 of the Design 3 interface changes will enter a new interface, which will be displayed as shown in Figure 11 of the Design 3 interface changes. Slide the page to display Figure 12 of the Design 3 interface changes. After sliding to a certain extent, it will display Figure 13 of the Design 3 interface changes, where you can select different models. After selecting a model, click "OK" in the upper right corner of Figure 13 of the Design 3 interface changes to enter Figure 14 of the Design 3 interface changes. Clicking "Submit Order" in Figure 14 of the Design 3 interface changes will enter a new interface, which will be displayed as shown in Figure 15 of the Design 3 interface changes. Clicking on any product in Figure 15 of the Design 3 interface changes will enter a new interface, which will be displayed as shown in Figure 16 of the Design 3 interface changes, where you can view a large image of the product. Design 4's main view is the initial interface for entering the mini-program. Design 4's main view is the initial interface for entering the mini-program. Slide the main view of Design 4 from left to right to display Design 4 interface change state diagram 1. After sliding, a new interface is entered, displaying Design 4 interface change state diagram 2. Slide from left to right in Design 4 interface change state diagram 2 to display Design 4 interface change state diagram 3. After sliding, a new interface is entered, displaying Design 4 interface change state diagram 4. Slide from left to right in Design 4 interface change state diagram 4 to display Design 4 interface change state diagram 5. After sliding, a new interface is entered, displaying Design 4 interface change state diagram 6. Click "Product" at the bottom of Design 4 interface change state diagram 6 to enter a new interface, displaying the interface state shown in Design 4 interface change state diagram 7. First, click the box at the top left of Design 4 interface change state diagram 7 to enter a new interface, displaying Design 4 interface change state diagram 8. After selecting any option, click the box on the right side of Design 4 interface change state diagram 8 and select any option to display Design 4 interface change state diagram 9. In Design 4 interface change state diagram 9, click "Product". Confirm to enter a new interface, as shown in Design 4 Interface Change State Diagram 10; Click on any product in Design 4 Interface Change State Diagram 10 to enter a new interface, as shown in Design 4 Interface Change State Diagram 11; Click on the option below "Select Model" in Design 4 Interface Change State Diagram 11 to enter a new interface, as shown in Design 4 Interface Change State Diagram 12; Scroll the page to see Design 4 Interface Change State Diagram 13; after scrolling to a certain extent, see Design 4 Interface Change State Diagram 14, where you can select different models. After selecting a model, click "OK" in the upper right corner of Design 4 Interface Change State Diagram 14 to enter Design 4 Interface Change State Diagram 15; Click "Submit Order" in Design 4 Interface Change State Diagram 15 to enter a new interface, as shown in Design 4 Interface Change State Diagram 16; Click on any product in Design 4 Interface Change State Diagram 16 to enter a new interface, as shown in Design 4 Interface Change State Diagram 17, where you can view a larger image of the product.

The embodiment of the invention provides a waterline extraction method and device, equipment, a storage medium and a product. The method comprises the steps of sampling a to-be-processed image to obtain a plurality of first image areas; performing waterline extraction on each first image area, and determining at least one second image area according to a waterline extraction result of each first image area; and obtaining a target waterline existing in the to-be-processed image according to the waterline extraction result of each first image area and the waterline extraction result of each second image area. According to the embodiment of the invention, the first image area is obtained through sampling to carry out waterline coarse screening, and then the second image area is directionally determined based on the extraction result to complete detail completion, so that a redundant reasoning task of full-quantity image segmentation is omitted; and meanwhile, an asynchronous cooperation and resource dynamic allocation mechanism is matched, and key area reasoning is focused, so that the resource utilization rate and the reasoning proportion are improved, the completeness of waterline extraction is ensured, and the large graph reasoning efficiency is remarkably improved.

The application provides a power defect image simulation method using heterogeneous computing resource load balancing scheduling, wherein a defect simulation task of a large image is split into defect simulation subtasks of several small images through image segmentation; then a task scheduling simulator is designed according to a cluster state of the heterogeneous computing resources and a defect simulation subtask state, the task scheduling simulator is used to plan task scheduling so that resource consumption is minimum; according to the planned task scheduling, the defect simulation subtasks are distributed to corresponding heterogeneous computing resources, and after image simulation, the defect simulation subtasks are combined to obtain a final defect simulation image. Through the use of the heterogeneous resources to schedule the small image defect simulation subtasks obtained by division, the average completion time of the simulation tasks is minimum, the computing efficiency is improved, and the defect simulation effect of the high-resolution large image of the power grid is ensured.

The application relates to the technical field of data analysis, and provides a method and device for information correlation of offline videos, equipment and a storage medium. The method comprises the following steps: preprocessing single-set videos, obtaining a maximum face area graph and a maximum single-face original graph of each type of face in a key frame; performing global clustering based on the maximum face area graph and the maximum single-face original graph of each type of face extracted from all single-set videos; selecting a global target face-removing large graph and a global target maximum single-face original graph of each type of face from the global clustering result; and calling an external star face recognition service for identity recognition based on the global target face-removing large graph and / or the global target maximum single-face original graph, and associating each recognition object with real information and role information. The application performs identity recognition after hierarchical processing, key frame screening and representative image selection, can reduce the data volume of global processing, and improves the processing speed and accuracy.

The invention discloses a large graph division method and system for realizing simultaneous double equalization of vertexes and connecting edges according to equal point / edge ratio expansion, the vertexes are taken as basic expansion units, the minimization of the difference value between the point / edge ratio of a current target block and the point / edge ratio of an original graph is taken as an optimization target of an expansion process, gradual expansion is performed to form a target block, and the target block is divided into two parts. The core comprises three parts: selection of seed vertexes, expansion exchange operation, generation of target blocks and adjustment of a residual graph, and the growth speeds of the vertexes and connecting edges can be controlled at the same time in the division process, so that the target blocks are balanced at the vertexes and the connecting edges at the same time; therefore, the graph division quality is improved, and the overall execution efficiency of distributed graph calculation is remarkably improved. Extended optimization is carried out on the basis of a traditional vertex equalization division method, and a connected edge equalization constraint is superposed on an existing vertex equalization mechanism, so that the feasibility and potential of realizing point-edge double equalization are better achieved.

The invention provides a remote sensing basic model reasoning acceleration method based on a super-heterogeneous system, which can be applied to the technical field of intelligent processing of remote sensing image data. The method comprises the following steps: loading a remote sensing basic model, and performing intelligent quantification on the remote sensing basic model in combination with hardware characteristics of a neural network processor to obtain a quantized model; reading the initial remote sensing large image by utilizing a field-programmable gate array, and preprocessing the initial remote sensing large image by adopting a data pipeline parallel mode to obtain preprocessed data; based on the preprocessed data, reasoning is conducted through a quantized model, an initial reasoning result is obtained, and hardware instruction optimization and algorithm optimization are conducted on the quantized model in the reasoning process through the quantized model; and restoring the initial reasoning result to the initial remote sensing large image according to a slice sequence to obtain a final reasoning result. The reasoning speed of the remote sensing basic model is improved from the two aspects of hardware instruction optimization and algorithm optimization, and rapid reasoning on the VPX board card is achieved.

A computing system can generate and train a machine-learning model for risk assessment. The machine-learning model can be trained on semi-labelled graph data that may contain one or more isolated nodes generated from a tabularized data set. The computing system can use graph embeddings to compare pairs of nodes of the graph to determine a similarity between each pair of nodes. The similarity may be used to determine whether to create a synthetic edge between the pair of nodes. An additional hyperparameter may be used to tune the number of generated edges based on a desired graph density. The generated graph data may then be used to train a machine-learning model capable of generating a risk indicator for a target entity. Further, the risk indicators can be utilized to control the access by a target entity to an interactive computing environment for accessing services provided by one or more institutions.