Auto-STGCN: Autonomous Spatial-Temporal Graph Convolutional Network Search

Abstract

In recent years, many spatial-temporal graph convolutional network (STGCN) models are proposed to deal with the spatial-temporal network data forecasting problem. These STGCN models have their own advantages, i.e., each of them puts forward many effective operations and achieves good prediction results in the real applications. If users can effectively utilize and combine these excellent operations integrating the advantages of existing models, then they may obtain more effective STGCN models thus create greater value using existing work. However, they fail to do so due to the lack of domain knowledge, and there is lack of automated system to help users to achieve this goal. In this paper, we fill this gap and propose Auto-STGCN algorithm, which makes use of existing models to automatically explore high-performance STGCN model for specific scenarios. Specifically, we design Unified-STGCN framework, which summarizes the operations of existing architectures, and use parameters to control the usage and characteristic attributes of each operation, so as to realize the parameterized representation of the STGCN architecture and the reorganization and fusion of advantages. Then, we present Auto-STGCN, an optimization method based on reinforcement learning, to quickly search the parameter search space provided by Unified-STGCN, and generate optimal STGCN models automatically. Extensive experiments on real-world benchmark datasets show that our Auto-STGCN can find STGCN models superior to existing STGCN models used for search space construction, which demonstrates the effectiveness of our proposed method.

Network traffic prediction based on feature fusion spatio-temporal graph convolutional network

“…The model is able to effectively capture complex local spatio-temporal correlations through a well-designed spatio-temporal synchronization modeling mechanism. Since there is no unified framework to describe the design process of STGCN models, Chunnan Wang et al proposed a reinforcement learning-based optimization method, Auto-STGCN [15], to quickly search the parameter search space provided by the unified STGCN and automatically generate the optimal STGCN model. Ta, Xuxiang et al In order to achieve multi-step traffic condition prediction, they proposed an adaptive spatio-temporal graph neural network called Ada-STNet [16], which firstly derives the optimal graph structure guided by node attributes, and then captures complex spatio-temporal correlations through a specialized spatio-temporal convolutional architecture.…”

A spatial–temporal graph deep learning model for urban flood nowcasting leveraging heterogeneous community features

“…Discriminative spatio-temporal graph convolutional network (DSTGCN) were used for action recognition in to inner-class action distribution 59 . Wang et al 5 developed an auto-STGCN algorithm that facilitates the detection of the optimal STGCNs models automatically using a reinforcement learning technique 60 . An attention mechanism allows DL models to focus more on the useful parts of features 46 .…”

Deep Learning on Graphs: Directed Graphs, Edge Structures and Graph Estimation

“…Various temporal models have been employed to predict traffic conditions. The majority uses recursive neural networks to model the temporal behaviour of traffic and a graphbased convolution to model the spatial behaviour (Li et al, 2018a;Zhao et al, 2019;Li et al, 2019c;Wang et al, 2020;Li et al, 2022;Deng et al, 2022). Alternatively, some authors avoid techniques employing recurrent units, owing to their needing to be unrolled, by using temporal convolution (Yu, Yin, and Zhu, 2018;Ta et al, 2022).…”