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

Cognitive grammar suggests that the acquisition of language grammar isgrounded within visual structures. While grammar is an essential representationof natural language, it also exists ubiquitously in vision to represent thehierarchical part-whole structure. In this work, we study grounded grammarinduction of vision and language in a joint learning framework. Specifically,we present VLGrammar, a method that uses compound probabilistic context-freegrammars (compound PCFGs) to induce the language grammar and the image grammarsimultaneously. We propose a novel contrastive learning framework to guide thejoint learning of both modules. To provide a benchmark for the grounded grammarinduction task, we collect a large-scale dataset, \textsc{PartIt}, whichcontains human-written sentences that describe part-level semantics for 3Dobjects. Experiments on the \textsc{PartIt} dataset show that VLGrammaroutperforms all baselines in image grammar induction and language grammarinduction. The learned VLGrammar naturally benefits related downstream tasks.Specifically, it improves the image unsupervised clustering accuracy by 30\%,and performs well in image retrieval and text retrieval. Notably, the inducedgrammar shows superior generalizability by easily generalizing to unseencategories.

VLGrammar: Grounded Grammar Induction of Vision and Language