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

Pretrained Vision-Language Models (VLMs) have achieved remarkable performancein image retrieval from text. However, their performance drops drastically whenconfronted with linguistically complex texts that they struggle to comprehend.Inspired by the Divide-and-Conquer algorithm and dual-process theory, in thispaper, we regard linguistically complex texts as compound proposition textscomposed of multiple simple proposition sentences and propose an end-to-endNeural Divide-and-Conquer Reasoning framework, dubbed NDCR. It contains threemain components: 1) Divide: a proposition generator divides the compoundproposition text into simple proposition sentences and produces theircorresponding representations, 2) Conquer: a pretrained VLMs-basedvisual-linguistic interactor achieves the interaction between decomposedproposition sentences and images, 3) Combine: a neural-symbolic reasonercombines the above reasoning states to obtain the final solution via a neurallogic reasoning approach. According to the dual-process theory, thevisual-linguistic interactor and neural-symbolic reasoner could be regarded asanalogical reasoning System 1 and logical reasoning System 2. We conductextensive experiments on a challenging image retrieval from contextualdescriptions data set. Experimental results and analyses indicate NDCRsignificantly improves performance in the complex image-text reasoning problem.Code link: https://github.com/YunxinLi/NDCR.

A Neural Divide-and-Conquer Reasoning Framework for Image Retrieval from Linguistically Complex Text