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

Recently, the superiority of Transformer for long-term time seriesforecasting (LTSF) tasks has been challenged, particularly since recent workhas shown that simple models can outperform numerous Transformer-basedapproaches. This suggests that a notable gap remains in fully leveraging thepotential of Transformer in LTSF tasks. Consequently, this study investigateskey issues when applying Transformer to LTSF, encompassing aspects of temporalcontinuity, information density, and multi-channel relationships. We introducethe Placeholder-enhanced Technique (PET) to enhance the computationalefficiency and predictive accuracy of Transformer in LTSF tasks. Furthermore,we delve into the impact of larger patch strategies and channel interactionstrategies on Transformer's performance, specifically Long Sub-sequenceDivision (LSD) and Multi-channel Separation and Interaction (MSI). Thesestrategies collectively constitute a novel model termed PETformer. Extensiveexperiments have demonstrated that PETformer achieves state-of-the-artperformance on eight commonly used public datasets for LTSF, surpassing allexisting models. The insights and enhancement methodologies presented in thispaper serve as valuable reference points and sources of inspiration for futureresearch endeavors.

PETformer: Long-term Time Series Forecasting via Placeholder-enhanced Transformer