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 data play an important role in many applications and theiranalysis reveals crucial information for understanding the underlyingprocesses. Among the many time series learning tasks of great importance, wehere focus on semi-supervised learning based on a graph representation of thedata. Two main aspects are involved in this task. A suitable distance measureto evaluate the similarities between time series, and a learning method to makepredictions based on these distances. However, the relationship between the twoaspects has never been studied systematically in the context of graph-basedlearning. We describe four different distance measures, including (Soft) DTWand MPDist, a distance measure based on the Matrix Profile, as well as foursuccessful semi-supervised learning methods, including the graph Allen--Cahnmethod and a Graph Convolutional Neural Network. We then compare theperformance of the algorithms on binary classification data sets. In ourfindings we compare the chosen graph-based methods using all distance measuresand observe that the results vary strongly with respect to the accuracy. Aspredicted by the ``no free lunch'' theorem, no clear best combination to employin all cases is found. Our study provides a reproducible framework for futurework in the direction of semi-supervised learning for time series with a focuson graph representations.

An Empirical Study of Graph-Based Approaches for Semi-Supervised Time Series Classification