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 address the overlooked unbiasedness in existing long-tailed classificationmethods: we find that their overall improvement is mostly attributed to thebiased preference of tail over head, as the test distribution is assumed to bebalanced; however, when the test is as imbalanced as the long-tailed trainingdata -- let the test respect Zipf's law of nature -- the tail bias is no longerbeneficial overall because it hurts the head majorities. In this paper, wepropose Cross-Domain Empirical Risk Minimization (xERM) for training anunbiased model to achieve strong performances on both test distributions, whichempirically demonstrates that xERM fundamentally improves the classification bylearning better feature representation rather than the head vs. tail game.Based on causality, we further theoretically explain why xERM achievesunbiasedness: the bias caused by the domain selection is removed by adjustingthe empirical risks on the imbalanced domain and the balanced but unseendomain. Codes are available at https://github.com/BeierZhu/xERM.

Cross-Domain Empirical Risk Minimization for Unbiased Long-tailed Classification