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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Machine learning, with its advances in deep learning has shown greatpotential in analyzing time series. In many scenarios, however, additionalinformation that can potentially improve the predictions is available. This iscrucial for data that arise from e.g., sensor networks that contain informationabout sensor locations. Then, such spatial information can be exploited bymodeling it via graph structures, along with the sequential (time series)information. Recent advances in adapting deep learning to graphs have shownpotential in various tasks. However, these methods have not been adapted fortime series tasks to a great extent. Most attempts have essentiallyconsolidated around time series forecasting with small sequence lengths.Generally, these architectures are not well suited for regression orclassification tasks where the value to be predicted is not strictly dependingon the most recent values, but rather on the whole length of the time series.We propose TISER-GCN, a novel graph neural network architecture for processing,in particular, these long time series in a multivariate regression task. Ourproposed model is tested on two seismic datasets containing earthquakewaveforms, where the goal is to predict maximum intensity measurements ofground shaking at each seismic station. Our findings demonstrate promisingresults of our approach -- with an average MSE reduction of 16.3% - compared tothe best performing baselines. In addition, our approach matches the baselinescores by needing only half the input size. The results are discussed in depthwith an additional ablation study.

Graph Neural Networks for Multivariate Time Series Regression with Application to Seismic Data