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

Reasoning about a model's accuracy on a test sample from its confidence is acentral problem in machine learning, being connected to important applicationssuch as uncertainty representation, model selection, and exploration. Whilethese connections have been well-studied in the i.i.d. settings, distributionshifts pose significant challenges to the traditional methods. Therefore, modelcalibration and model selection remain challenging in the unsupervised domainadaptation problem--a scenario where the goal is to perform well in adistribution shifted domain without labels. In this work, we tackledifficulties coming from distribution shifts by developing a novel importanceweighted group accuracy estimator. Specifically, we formulate an optimizationproblem for finding an importance weight that leads to an accurate groupaccuracy estimation in the distribution shifted domain with theoreticalanalyses. Extensive experiments show the effectiveness of group accuracyestimation on model calibration and model selection. Our results emphasize thesignificance of group accuracy estimation for addressing challenges inunsupervised domain adaptation, as an orthogonal improvement direction withimproving transferability of accuracy.

IW-GAE: Importance weighted group accuracy estimation for improved calibration and model selection in unsupervised domain adaptation