ST-UNet: A Spatio-Temporal U-Network for Graph-structured Time Series Modeling

Abstract

The spatio-temporal graph learning is becoming an increasingly important object of graph study. Many application domains involve highly dynamic graphs where temporal information is crucial, e.g. traffic networks and financial transaction graphs. Despite the constant progress made on learning structured data, there is still a lack of effective means to extract dynamic complex features from spatio-temporal structures. Particularly, conventional models such as convolutional networks or recurrent neural networks are incapable of revealing the temporal patterns in short or long terms and exploring the spatial properties in local or global scope from spatiotemporal graphs simultaneously. To tackle this problem, we design a novel multi-scale architecture, Spatio-Temporal U-Net (ST-UNet), for graph-structured time series modeling. In this U-shaped network, a paired sampling operation is proposed in spacetime domain accordingly: the pooling (ST-Pool) coarsens the input graph in spatial from its deterministic partition while abstracts multi-resolution temporal dependencies through dilated recurrent skip connections; based on previous settings in the downsampling, the unpooling (ST-Unpool) restores the original structure of spatio-temporal graphs and resumes regular intervals within graph sequences. Experiments on spatio-temporal prediction tasks demonstrate that our model effectively captures comprehensive features in multiple scales and achieves substantial improvements over mainstream methods on several real-world datasets.

A Hybrid Model for Spatiotemporal Air Quality Prediction Based on Interpretable Neural Networks and a Graph Neural Network

“…ST-UNet [65]: ST-UNet is a neural network model for processing graph-structured time series data. To improve the time series modeling ability of the graph-structured time series data, the model presented an the extended GRU.…”

Thermodynamics-informed graph neural networks

“…Future work may overcome the stated limitations by combining graph representations with model order reduction techniques, such as autoencoders or U-net architectures [42,43]. The idea is to replace deep message passing with various coarse-graining steps, allowing the boundary information to reach every node in the simulation domain while reducing the number of parameters to the network.…”

Efficient Colon Cancer Grading with Graph Neural Networks

“…Close to this work, Zhou et al (2019) proposed a graph neural network that has a three layered approach of their adaptive GraphSage model, which contains a LSTM structure for information aggregation over three different graph convolutional layers. Graph neural networks have also been used for other medical applications, publications from Yu et al (2019) andJuarez et al (2019) propose to use graph neural network for 2D and or 3D image analysis of medical data, mostly scans. Related to this work in terms of graph neural network theory and architectures, various authors have provided detailed benchmarks on aggregation methods and graph neural network architectures tested over common graph data sets such as biomedical graphs or social graphs, see Dwivedi et al (2020), Morris et al (2019) and Xu et al (2018).…”