Multi-entity data stream classification method based on space-time attention graph neural network

• Main Authors: ZUO QIONG, QIAN LIPENG, SUN GUANQUN, LEE DAE-HO, ZHU HONG, WANG ERXI, CAO ZHONGSHENG
• Format: Patent
• Language: Chinese; English
• Published: 05.07.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a multi-entity data stream classification method based on a space-time attention graph neural network, and belongs to the field of multi-entity data stream classification, and the method comprises the steps: obtaining data of each entity at a t moment as an input data sequence, and reconstructing the input data sequence into a graph structure Gt; when t is equal to 1, performing multiple times of message passing reconstruction on each node feature in the graph structure G1 by adopting a space attention mechanism to obtain a space correlation feature of each node and form a graph structure G '1; when t is greater than or equal to 2, carrying out information aggregation on the graph structure Gt and the graph structure G't-1 by adopting a time attention mechanism to obtain a graph structure G ''t, and carrying out multiple times of message passing reconstruction on each node feature in the graph structure G' 't by adopting a space attention mechanism to obtain a space correlation feature