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

Multivariable time series classification problems are increasing inprevalence and complexity in a variety of domains, such as biology and finance.While deep learning methods are an effective tool for these problems, theyoften lack interpretability. In this work, we propose a novel modular prototypelearning framework for multivariable time series classification. In the firststage of our framework, encoders extract features from each variableindependently. Prototype layers identify single-variable prototypes in theresulting feature spaces. The next stage of our framework represents themultivariable time series sample points in terms of their similarity to thesesingle-variable prototypes. This results in an inherently interpretablerepresentation of multivariable patterns, on which prototype learning isapplied to extract representative examples i.e. multivariable prototypes. Ourframework is thus able to explicitly identify both informative patterns in theindividual variables, as well as the relationships between the variables. Wevalidate our framework on a simulated dataset with embedded patterns, as wellas a real human activity recognition problem. Our framework attains comparableor superior classification performance to existing time series classificationmethods on these tasks. On the simulated dataset, we find that our modelreturns interpretations consistent with the embedded patterns. Moreover, theinterpretations learned on the activity recognition dataset align with domainknowledge.

Multi-Modal Prototype Learning for Interpretable Multivariable Time Series Classification