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

BERT-based models have had strong performance on leaderboards, yet have beendemonstrably worse in real-world settings requiring generalization. Limitedquantities of training data is considered a key impediment to achievinggeneralizability in machine learning. In this paper, we examine the impact oftraining data quality, not quantity, on a model's generalizability. We considertwo characteristics of training data: the portion of human-adversarial(h-adversarial), i.e., sample pairs with seemingly minor differences butdifferent ground-truth labels, and human-affable (h-affable) training samples,i.e., sample pairs with minor differences but the same ground-truth label. Wefind that for a fixed size of training samples, as a rule of thumb, having10-30% h-adversarial instances improves the precision, and therefore F1, by upto 20 points in the tasks of text classification and relation extraction.Increasing h-adversarials beyond this range can result in performance plateausor even degradation. In contrast, h-affables may not contribute to a model'sgeneralizability and may even degrade generalization performance.

Effects of Human Adversarial and Affable Samples on BERT Generalization