7results about How to "Accurate aggregation" patented technology

The invention provides a multi-task federated learning client contribution degree measurement method based on model layering, and relates to a federated learning and machine learning model layering technology. According to the method, a distributed architecture of a central server and a plurality of client nodes is included, firstly, a federation learning environment and data preparation are configured, then, the central server measures the contribution degree of each client under an existing federation task, and then, the central server layers received model parameters and calculates the gradient of each layer of a model; the model layering similarity relative to the initiator client is calculated to obtain a normalized contribution vector, a new model is obtained through calculation according to the contribution vector and the data size on the client, the client participating in training in the next round is selected, and the task queue is updated. According to the method, the contribution of each client of federal learning can be accurately identified, more accurate model aggregation under layered contribution measurement is realized, meanwhile, the privacy protection effect the same as that of general federal learning is kept, the federal learning time is optimized, and the model precision is improved.

This application provides a network channel routing verification method and apparatus. The method constructs a directed graph based on the directed connection relationships of ports in the routing data, and uses graph algorithms to partition connected components and add audit rules based on data dimensions for verification. This allows for rapid and accurate detection of the connectivity and integrity of channel routes. During the verification process, it does not rely on the segment group order attribute in the routing data, avoiding shallow data quality issues that could prevent verification. Furthermore, it can verify the integrity of the channel carrying structure according to the business channel dimension, quickly aggregating all hierarchical channel routing anomaly audit results. For business channels with abnormal aggregated audit results, a secondary verification is performed to ensure the accuracy and reliability of the verification results. This helps vendors rectify data according to the business channel dimension and achieve rapid results.

The application relates to a power grid risk detection method, device and equipment based on a knowledge graph. The method comprises the following steps: acquiring multi-source detection data of a power grid to be detected; performing feature extraction on the multi-source detection data to obtain a plurality of feature data; performing mapping processing on the feature data and a preset knowledge graph according to the label of each feature data, and determining a target knowledge graph comprising the plurality of feature data according to the mapping result; wherein the preset knowledge graph is used for representing various types of preset entities, preset attributes of each type of preset entity, and constraint relationships between the preset entities; clustering target entities corresponding to the feature data with a category label according to the target knowledge graph to obtain a plurality of feature data clusters; performing risk detection on each feature data cluster, and determining a risk detection result of the power grid to be detected according to the detection result. The method can improve the accuracy of power grid risk detection.

The invention belongs to the technical field of crop variety innovation, and particularly relates to a breeding method and application of wheat capable of resisting gibberellic disease and herbicide-tolerant starter propagation. According to the method disclosed by the invention, through combination of molecular marker-assisted selection and backcross breeding, excellent characters from different donors are rapidly introduced into a genetic background of an excellent koji-making variety, so that accurate and efficient polymerization of target characters is realized, and the screening efficiency and accuracy can be improved; meanwhile, the screening index of starter propagation quality runs through the breeding process, so that the bred material is not only excellent in resistance and agronomic character, but also can meet the specific requirements of the terminal industry. The variety bred by the method provided by the invention can reduce the use of pesticides for preventing and treating gibberellic disease and reduce the pollution risk of fungaltoxin; the herbicide-resistant characteristic facilitates field weed management, and the yield and quality stability are improved; the special quality guarantees the high additional value in the starter propagation industry, and creates the value of the whole industrial chain from the field to the workshop.

This invention provides a spherical screen support structure system, belonging to the field of radio frequency simulation system technology. It includes a support frame, a load-bearing structure, multiple horizontal support rods, and multiple inclined support rods. The support frame is arranged in an arc around the spherical screen, and the load-bearing structure is mounted on the support frame. One side of the load-bearing structure protrudes from the spherical surface, and the other side has multiple first mounting cylinders, with the central axes of the multiple first mounting cylinders converging at the center of the protruding spherical surface. Horizontal support rods are installed between two adjacent left and right first mounting cylinders, and inclined support rods are installed between two adjacent vertically mounted first mounting cylinders. The three adjacent first mounting cylinders, one horizontal support rod, and two inclined support rods are arranged in a triangle to form a ternary support component. The spherical screen support structure system provided by this invention can effectively control the positional accuracy of each antenna, significantly improving the performance and accuracy of the simulation system; it also enhances the installation stability of the antenna units.

The invention relates to the technical field of flow aggregation, in particular to a multistage homologous session flow aggregation method and system, and the method comprises the steps: extracting quintuple information of each data packet, generating a session identifier, and calculating a session feature vector and a session priority score; at the first aggregation level, each service interface board establishes a local session table; preferentially performing aggregation processing on the data packets of the high-priority sessions according to the session priority scores; at a second aggregation level, each service interface board synchronizes the local session table to a case-level session table; forwarding the data packets with the same session identifier to the same output port; at a third aggregation level, synchronizing the case-level session table to a global session table; redirecting the data packets on other cases to a corresponding output port of the main aggregation node determined according to the session affinity; outputting all the data packets of the same session identifier from the same output port and generating a session integrity report; therefore, the integrity and consistency of the session data are ensured.

The invention relates to the technical field of filter media, in particular to an acid-resistant nanofiltration membrane and a preparation method thereof.The preparation method sequentially comprises the following steps that S1, a water phase solution and an oil phase solution are provided; the aqueous phase solution is an aqueous solution in which a polyamine monomer is dissolved; the oil phase solution is an organic solution in which an isocyanate monomer and an alkyl phosphate salt compound serving as a reaction accelerator are dispersed; s2, enabling the water-phase solution and the oil-phase solution to be in contact on a supporting base membrane to generate interfacial polymerization reaction, so as to form a polyurea active separation layer on the surface of the supporting base membrane; s3, performing heat treatment to obtain the strong acid-resistant nanofiltration membrane. According to the invention, a small amount of alkyl phosphate salt compound is added into the oil-phase solution as a reaction accelerator, so that the formation of a polyurea active separation layer with a better structure can be promoted, and the water flux, the acid transmittance and the metal ion rejection rate of the nanofiltration membrane are synchronously improved.