A real-time graph sampling method based on a flooding broadcast GNN network and a terminal
By using a flooding broadcast GNN network to perform multiple flooding and neighbor sampling on dynamic images, the latency and noise problems in large-scale image processing are solved, and accurate sampling of dynamic images is achieved.
CN118710494BActive Publication Date: 2025-12-12FUJIAN TIANQUAN EDUCATION TECH LTD
Patent Information
- Application Number
- CN202410817072.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-24
- Publication Date
- 2025-12-12
- Estimated Expiration
- 2044-06-24
Smart Images
Figure CN118710494B_ABST
Abstract
The application provides a real-time graph sampling method and a terminal based on a flooding broadcast GNN network, which comprises the following steps: inputting a real-time dynamic image into the GNN network, and performing initial flooding broadcast to obtain initial flooding data; performing secondary flooding broadcast on the initial flooding data to obtain secondary flooding data; performing neighbor sampling on the same data in the secondary flooding data as the initial flooding data, returning the different data to step S2 to perform secondary flooding broadcast again, and storing the data in different sets respectively; and repeating the above steps until the dynamic image is inputted completely, splicing the data in each set according to a time stamp, and outputting a sampling result. The application effectively realizes the purpose of accurate sampling of real-time dynamic images.
Need to check novelty before this filing date? Find Prior Art
Citation Information
Patent Citations
Graph attention network inductive learning method based on graph sampling
CN113077003A
Fixed-length side point combined sampling mechanism in graph neural network training
CN116841762A