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

Word translation without parallel corpora has become feasible, rivaling theperformance of supervised methods. Recent findings have shown that the accuracyand robustness of unsupervised word translation (UWT) can be improved by makinguse of visual observations, which are universal representations acrosslanguages. In this work, we investigate the potential of using not only visualobservations but also pretrained language-image models for enabling a moreefficient and robust UWT. Specifically, we develop a novel UWT method dubbedWord Alignment using Language-Image Pretraining (WALIP), which leverages visualobservations via the shared embedding space of images and texts provided byCLIP models (Radford et al., 2021). WALIP has a two-step procedure. First, weretrieve word pairs with high confidences of similarity, computed using ourproposed image-based fingerprints, which define the initial pivot for the wordalignment. Second, we apply our robust Procrustes algorithm to estimate thelinear mapping between two embedding spaces, which iteratively corrects andrefines the estimated alignment. Our extensive experiments show that WALIPimproves upon the state-of-the-art performance of bilingual word alignment fora few language pairs across different word embeddings and displays greatrobustness to the dissimilarity of language pairs or training corpora for twoword embeddings.

Utilizing Language-Image Pretraining for Efficient and Robust Bilingual Word Alignment