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

When dealing with binary classification of data with only one labeled classdata scientists employ two main approaches, namely One-Class (OC)classification and Positive Unlabeled (PU) learning. The former only learnsfrom labeled positive data, whereas the latter also utilizes unlabeled data toimprove the overall performance. Since PU learning utilizes more data, we mightbe prone to think that when unlabeled data is available, the go-to algorithmsshould always come from the PU group. However, we find that this is not alwaysthe case if unlabeled data is unreliable, i.e. contains limited or biasedlatent negative data. We perform an extensive experimental study of a wide listof state-of-the-art OC and PU algorithms in various scenarios as far asunlabeled data reliability is concerned. Furthermore, we propose PUmodifications of state-of-the-art OC algorithms that are robust to unreliableunlabeled data, as well as a guideline to similarly modify other OC algorithms.Our main practical recommendation is to use state-of-the-art PU algorithms whenunlabeled data is reliable and to use the proposed modifications ofstate-of-the-art OC algorithms otherwise. Additionally, we outline proceduresto distinguish the cases of reliable and unreliable unlabeled data usingstatistical tests.

Improving State-of-the-Art in One-Class Classification by Leveraging Unlabeled Data