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

We propose Okapi, a simple, efficient, and general method for robustsemi-supervised learning based on online statistical matching. Our method usesa nearest-neighbours-based matching procedure to generate cross-domain viewsfor a consistency loss, while eliminating statistical outliers. In order toperform the online matching in a runtime- and memory-efficient way, we drawupon the self-supervised literature and combine a memory bank with aslow-moving momentum encoder. The consistency loss is applied within thefeature space, rather than on the predictive distribution, making the methodagnostic to both the modality and the task in question. We experiment on theWILDS 2.0 datasets Sagawa et al., which significantly expands the range ofmodalities, applications, and shifts available for studying and benchmarkingreal-world unsupervised adaptation. Contrary to Sagawa et al., we show that itis in fact possible to leverage additional unlabelled data to improve uponempirical risk minimisation (ERM) results with the right method. Our methodoutperforms the baseline methods in terms of out-of-distribution (OOD)generalisation on the iWildCam (a multi-class classification task) andPovertyMap (a regression task) image datasets as well as the CivilComments (abinary classification task) text dataset. Furthermore, from a qualitativeperspective, we show the matches obtained from the learned encoder are stronglysemantically related. Code for our paper is publicly available athttps://github.com/wearepal/okapi/.

Okapi: Generalising Better by Making Statistical Matches Match