English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Graph neural networks (GNNs), especially dynamic GNNs, have become a researchhotspot in spatio-temporal forecasting problems. While many dynamic graphconstruction methods have been developed, relatively few of them explore thecausal relationship between neighbour nodes. Thus, the resulting models lackstrong explainability for the causal relationship between the neighbour nodesof the dynamically generated graphs, which can easily lead to a risk insubsequent decisions. Moreover, few of them consider the uncertainty and noiseof dynamic graphs based on the time series datasets, which are ubiquitous inreal-world graph structure networks. In this paper, we propose a novel DynamicDiffusion-Variational Graph Neural Network (DVGNN) for spatio-temporalforecasting. For dynamic graph construction, an unsupervised generative modelis devised. Two layers of graph convolutional network (GCN) are applied tocalculate the posterior distribution of the latent node embeddings in theencoder stage. Then, a diffusion model is used to infer the dynamic linkprobability and reconstruct causal graphs in the decoder stage adaptively. Thenew loss function is derived theoretically, and the reparameterization trick isadopted in estimating the probability distribution of the dynamic graphs byEvidence Lower Bound during the backpropagation period. After obtaining thegenerated graphs, dynamic GCN and temporal attention are applied to predictfuture states. Experiments are conducted on four real-world datasets ofdifferent graph structures in different domains. The results demonstrate thatthe proposed DVGNN model outperforms state-of-the-art approaches and achievesoutstanding Root Mean Squared Error result while exhibiting higher robustness.Also, by F1-score and probability distribution analysis, we demonstrate thatDVGNN better reflects the causal relationship and uncertainty of dynamicgraphs.

Dynamic Causal Explanation Based Diffusion-Variational Graph Neural Network for Spatio-temporal Forecasting