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

Most machine learning methods are used as a black box for modelling. We maytry to extract some knowledge from physics-based training methods, such asneural ODE (ordinary differential equation). Neural ODE has advantages like apossibly higher class of represented functions, the extended interpretabilitycompared to black-box machine learning models, ability to describe both trendand local behaviour. Such advantages are especially critical for time serieswith complicated trends. However, the known drawback is the high training timecompared to the autoregressive models and long-short term memory (LSTM)networks widely used for data-driven time series modelling. Therefore, weshould be able to balance interpretability and training time to apply neuralODE in practice. The paper shows that modern neural ODE cannot be reduced tosimpler models for time-series modelling applications. The complexity of neuralODE is compared to or exceeds the conventional time-series modelling tools. Theonly interpretation that could be extracted is the eigenspace of the operator,which is an ill-posed problem for a large system. Spectra could be extractedusing different classical analysis methods that do not have the drawback ofextended time. Consequently, we reduce the neural ODE to a simpler linear formand propose a new view on time-series modelling using combined neural networksand an ODE system approach.

On the balance between the training time and interpretability of neural ODE for time series modelling