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

Transfer learning is beneficial by allowing the expressive features of modelspretrained on large-scale datasets to be finetuned for the target task ofsmaller, more domain-specific datasets. However, there is a concern that thesepretrained models may come with their own biases which would propagate into thefinetuned model. In this work, we investigate bias when conceptualized as bothspurious correlations between the target task and a sensitive attribute as wellas underrepresentation of a particular group in the dataset. Under both notionsof bias, we find that (1) models finetuned on top of pretrained models canindeed inherit their biases, but (2) this bias can be corrected for throughrelatively minor interventions to the finetuning dataset, and often with anegligible impact to performance. Our findings imply that careful curation ofthe finetuning dataset is important for reducing biases on a downstream task,and doing so can even compensate for bias in the pretrained model.

Overwriting Pretrained Bias with Finetuning Data