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

We aim to dismantle the prevalent black-box neural architectures used incomplex visual reasoning tasks, into the proposed eXplainable and eXplicitNeural Modules (XNMs), which advance beyond existing neural module networkstowards using scene graphs --- objects as nodes and the pairwise relationshipsas edges --- for explainable and explicit reasoning with structured knowledge.XNMs allow us to pay more attention to teach machines how to "think",regardless of what they "look". As we will show in the paper, by using scenegraphs as an inductive bias, 1) we can design XNMs in a concise and flexiblefashion, i.e., XNMs merely consist of 4 meta-types, which significantly reducethe number of parameters by 10 to 100 times, and 2) we can explicitly trace thereasoning-flow in terms of graph attentions. XNMs are so generic that theysupport a wide range of scene graph implementations with various qualities. Forexample, when the graphs are detected perfectly, XNMs achieve 100% accuracy onboth CLEVR and CLEVR CoGenT, establishing an empirical performance upper-boundfor visual reasoning; when the graphs are noisily detected from real-worldimages, XNMs are still robust to achieve a competitive 67.5% accuracy onVQAv2.0, surpassing the popular bag-of-objects attention models without graphstructures.

Explainable and Explicit Visual Reasoning over Scene Graphs