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

Image and text retrieval is one of the foundational tasks in the vision andlanguage domain with multiple real-world applications. State-of-the-artapproaches, e.g. CLIP, ALIGN, represent images and texts as dense embeddingsand calculate the similarity in the dense embedding space as the matchingscore. On the other hand, sparse semantic features like bag-of-words models aremore interpretable, but believed to suffer from inferior accuracy than denserepresentations. In this work, we show that it is possible to build a sparsesemantic representation that is as powerful as, or even better than, densepresentations. We extend the CLIP model and build a sparse text and imagerepresentation (STAIR), where the image and text are mapped to a sparse tokenspace. Each token in the space is a (sub-)word in the vocabulary, which is notonly interpretable but also easy to integrate with existing informationretrieval systems. STAIR model significantly outperforms a CLIP model with+$4.9\%$ and +$4.3\%$ absolute Recall@1 improvement on COCO-5ktext$\rightarrow$image and image$\rightarrow$text retrieval respectively. Italso achieved better performance on both of ImageNet zero-shot and linearprobing compared to CLIP.

STAIR: Learning Sparse Text and Image Representation in Grounded Tokens