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

Time series forecasting has widespread applications in urban life rangingfrom air quality monitoring to traffic analysis. However, accurate time seriesforecasting is challenging because real-world time series suffer from thedistribution shift problem, where their statistical properties change overtime. Despite extensive solutions to distribution shifts in domain adaptationor generalization, they fail to function effectively in unknown,constantly-changing distribution shifts, which are common in time series. Inthis paper, we propose Hyper Time- Series Forecasting (HTSF), ahypernetwork-based framework for accurate time series forecasting underdistribution shift. HTSF jointly learns the time-varying distributions and thecorresponding forecasting models in an end-to-end fashion. Specifically, HTSFexploits the hyper layers to learn the best characterization of thedistribution shifts, generating the model parameters for the main layers tomake accurate predictions. We implement HTSF as an extensible framework thatcan incorporate diverse time series forecasting models such as RNNs andTransformers. Extensive experiments on 9 benchmarks demonstrate that HTSFachieves state-of-the-art performances.

Combating Distribution Shift for Accurate Time Series Forecasting via Hypernetworks