7results about How to "Improve trustworthiness" patented technology

The embodiment of the invention provides a user risk classification model training method and device and a request response method and device. The specific implementation scheme of the training method comprises the following steps: obtaining each sample behavior feature of a sample user and a risk category label of the sample user; performing feature fusion on each sample behavior feature of the sample user to obtain a user feature of the sample user; inputting the user features of the sample user into a to-be-trained user risk classification model to obtain a prediction result of the sample user; determining a loss value of the user risk classification model based on the prediction result of the sample user and the risk category label of the sample user; and when judging that the user risk classification model is not converged based on the loss value, adjusting network parameters of the user risk classification model. Therefore, according to the training method, the model capable of effectively identifying the user risk category can be obtained through training, and an implementation basis is provided for effectively distinguishing the risk category of the user.

The present invention provides a display panel and a manufacturing method thereof and a display device. The display panel comprises a display backplane and a light-emitting element, wherein the display backplane comprises a substrate and a first electrode, wherein the first electrode is arranged on the surface of the substrate; the light-emitting element comprises a light-emitting body and a second electrode, wherein the second electrode is arranged on the surface of the light-emitting body and the second electrode is electrically connected to the light-emitting body; the first electrode and the second electrode are in contact and electrically connected, wherein the first electrode comprises a protruding structure, wherein the protruding structure is arranged on the contact surface of the first electrode with the second electrode; and / or the second electrode comprises a protruding structure, wherein the protruding structure is arranged on the contact surface of the second electrode with the first electrode. The display panel can absorb and restrict the molten solder, reduce its diffusion area, avoid a large diffusion range after the solder is melted, and avoid a short circuit of the light-emitting element; and is conducive to the release of gas generated during welding, reduces the pores generated in the solder during welding, and avoids a short circuit of the light-emitting element.

This invention belongs to the technical field of computer-aided design and simulation, and relates to a method and system for calculating the leakage of hydraulic facilities by integrating multi-source data. The method includes: receiving raw sensor time-series signals and reconstructing a multi-dimensional time evolution trajectory in a local phase space; extracting the dynamic deviation at abrupt changes and encoding it as a hydraulic transient characteristic tensor; performing singular value decomposition on the tensor using a discrete fractured mesh model to extract the feature spectrum; when it matches the critical deformation matrix, activating the nonlinear turbulence equation to rewrite the state parameters and outputting the pressure head distribution; calculating the simulated entropy yield based on the flow field gradient characteristics and thermodynamic laws; if it is less than a compliance threshold, generating a channel mask array to shield abnormal data sources; if the threshold is met, calculating the facility leakage. This invention solves the problem that insufficient fusion depth between the physical model and multi-source monitoring data leads to limitations in the reliability and real-time response of leakage calculation under complex operating conditions.

This application discloses a semiconductor structure and its manufacturing method. The semiconductor structure is axisymmetric and includes: an active region; a recessed portion formed in the center of the active region, and a protrusion formed on each side of the recessed portion; a drain located at the bottom of the recessed portion; a data line contact located on the drain; a gate sidewall layer located on one side of the data line contact; two isolation structures located on both sides of the active region; two gate layers located on the side of the gate sidewall layer on the same side away from the axis of symmetry; two word lines located above the gate layers on the same side; and two sources located on the protrusions on the same side. The semiconductor structure of this application can reduce interference errors caused by inter-cell interference due to miniaturization of the manufacturing process, prevent characteristic changes (degradation) caused by inter-row coupling, and effectively improve the reliability of highly integrated semiconductor products.

The invention provides a method for loading a three-dimensional model during operation based on UE5, and relates to the technical field of computer systems.The method for loading the three-dimensional model comprises an asset management module, a loading module and a synchronizing module, downloading operation is executed through an asynchronous task by means of the asset management module, the blocking phenomenon of a main thread and a rendering thread is remarkably reduced, and the loading efficiency is improved. Therefore, the overall response speed and the operation smoothness of the system are improved. Hash calculation and a database query mechanism are utilized to efficiently judge whether the file exists or not, repeated downloading is avoided, and bandwidth and storage resources are saved. Besides, due to the application of the multi-thread and parallel processing technology, downloading, loading and synchronization tasks can be executed in parallel, the performance of a multi-core processor is fully utilized, and the resource utilization efficiency and the processing capacity of the system are further improved. The optimization measures ensure that the user can obtain seamless and rapid loading experience in the using process, and the practicability of the system and the user satisfaction degree are greatly improved.

The invention discloses a laboratory quality control data management method, and relates to the technical field of big data management, and the method comprises the steps: obtaining laboratory quality control data, and carrying out the classification according to the food category, and obtaining a plurality of quality control data sets; performing trend consistency analysis on a plurality of pieces of detection category data in the quality control data sequence to obtain category credibility; performing category integrity analysis on the quality control data sequence to obtain integrity credibility; screening to obtain quality control data sequences of a plurality of same storage places, and performing transverse trend consistency analysis to obtain transverse credibility; and performing weighted calculation on the category credibility, the integrity credibility and the transverse credibility to obtain comprehensive credibility, performing credibility grading on the quality control data sequence, performing hierarchical storage, and performing laboratory quality control data management. The technical problem of poor quality control data management effect in the prior art is solved.

The invention belongs to the technical field of artificial intelligence and unmanned systems, and particularly discloses an unmanned system-oriented dual-channel knowledge-driven network trajectory prediction method, which comprises the following steps of: receiving scene data of a dynamic target in an operation scene of an unmanned system, and constructing a dynamic heterogeneous graph comprising a plurality of semantic nodes and a plurality of connection relationships based on the scene data; a two-channel encoder is adopted to encode the dynamic heterogeneous graph, and a first feature sequence representing dynamic interaction and a second feature sequence representing instantaneous physical geometry are obtained; the dual-channel encoder comprises a knowledge-driven dynamic interaction channel and a parallel instantaneous physical geometry channel; fusing the features of the first feature sequence and the second sequence by adopting a dynamic gating mechanism to obtain a unified scene context vector; and decoding the scene context vector by adopting a query-based multi-mode decoder to generate at least one group of prediction tracks and determine confidence scores corresponding to the prediction tracks. According to the invention, the accuracy and reliability of trajectory prediction can be improved.