Insulator defect recognition based on cross-level feature fusion and channel and spatial attention mechanism
This paper addresses the problem of real-time traffic prediction in Intelligent Transportation Systems(ITS)and travel navigation guidance.It proposes a Dynamic Temporal Self-Attention Graph Convolutional Network(DT-SGN)model.Existing traffic prediction methods often use a simple adjacency matrix consisting of 0 and 1 to capture spatial dependencies,which cannot accurately describe the topological structure of urban road networks and their temporal dynamics.To tackle this issue,this paper treats the adjacency matrix as a trainable attention score matrix and adapts network parameters to different inputs.Specifically,the Self-Attention Graph Convolutional Network(SGN)is employed to capture spatial dependencies,while the Dynamic Gated Recurrent Unit(DGRU)is utilized to capture temporal dependencies and learn the dynamic changes in input data.Experimental results on the SZ-taxi and Los-loop dataset demonstrate the superiority of our proposed method over model-driven and data-driven comparative approaches when applied to real-world traffic datasets.
graph networkdeep learningtraffic predictiontime and spatial mechanisms
NETL
NSTL
万方数据
