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

Human similarity judgments are a powerful supervision signal for machinelearning applications based on techniques such as contrastive learning,information retrieval, and model alignment, but classical methods forcollecting human similarity judgments are too expensive to be used at scale.Recent methods propose using pre-trained deep neural networks (DNNs) toapproximate human similarity, but pre-trained DNNs may not be available forcertain domains (e.g., medical images, low-resource languages) and theirperformance in approximating human similarity has not been extensively tested.We conducted an evaluation of 611 pre-trained models across three domains --images, audio, video -- and found that there is a large gap in performancebetween human similarity judgments and pre-trained DNNs. To address this gap,we propose a new class of similarity approximation methods based on language.To collect the language data required by these new methods, we also developedand validated a novel adaptive tag collection pipeline. We find that ourproposed language-based methods are significantly cheaper, in the number ofhuman judgments, than classical methods, but still improve performance over theDNN-based methods. Finally, we also develop `stacked' methods that combinelanguage embeddings with DNN embeddings, and find that these consistentlyprovide the best approximations for human similarity across all three of ourmodalities. Based on the results of this comprehensive study, we provide aconcise guide for researchers interested in collecting or approximating humansimilarity data. To accompany this guide, we also release all of the similarityand language data, a total of 206,339 human judgments, that we collected in ourexperiments, along with a detailed breakdown of all modeling results.

Words are all you need? Language as an approximation for human similarity judgments