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

Anomaly detection attempts at finding examples that deviate from the expectedbehaviour. Usually, anomaly detection is tackled from an unsupervisedperspective because anomalous labels are rare and difficult to acquire.However, the lack of labels makes the anomaly detector have high uncertainty insome regions, which usually results in poor predictive performance or low usertrust in the predictions. One can reduce such uncertainty by collectingspecific labels using Active Learning (AL), which targets examples close to thedetector's decision boundary. Alternatively, one can increase the user trust byallowing the detector to abstain from making highly uncertain predictions,which is called Learning to Reject (LR). One way to do this is by thresholdingthe detector's uncertainty based on where its performance is low, whichrequires labels to be evaluated. Although both AL and LR need labels, they workwith different types of labels: AL seeks strategic labels, which are evidentlybiased, while LR requires i.i.d. labels to evaluate the detector's performanceand set the rejection threshold. Because one usually has a unique label budget,deciding how to optimally allocate it is challenging. In this paper, we proposea mixed strategy that, given a budget of labels, decides in multiple roundswhether to use the budget to collect AL labels or LR labels. The strategy isbased on a reward function that measures the expected gain when allocating thebudget to either side. We evaluate our strategy on 18 benchmark datasets andcompare it to some baselines.

How to Allocate your Label Budget? Choosing between Active Learning and Learning to Reject in Anomaly Detection