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

Reasoning is a critical ability towards complete visual understanding. Todevelop machine with cognition-level visual understanding and reasoningabilities, the visual commonsense reasoning (VCR) task has been introduced. InVCR, given a challenging question about an image, a machine must answercorrectly and then provide a rationale justifying its answer. The methodsadopting the powerful BERT model as the backbone for learning jointrepresentation of image content and natural language have shown promisingimprovements on VCR. However, none of the existing methods have utilizedcommonsense knowledge in visual commonsense reasoning, which we believe will begreatly helpful in this task. With the support of commonsense knowledge,complex questions even if the required information is not depicted in the imagecan be answered with cognitive reasoning. Therefore, we incorporate commonsenseknowledge into the cross-modal BERT, and propose a novel Knowledge EnhancedVisual-and-Linguistic BERT (KVL-BERT for short) model. Besides taking visualand linguistic contents as input, external commonsense knowledge extracted fromConceptNet is integrated into the multi-layer Transformer. In order to reservethe structural information and semantic representation of the originalsentence, we propose using relative position embedding and mask-self-attentionto weaken the effect between the injected commonsense knowledge and otherunrelated components in the input sequence. Compared to other task-specificmodels and general task-agnostic pre-training models, our KVL-BERT outperformsthem by a large margin.

KVL-BERT: Knowledge Enhanced Visual-and-Linguistic BERT for Visual Commonsense Reasoning