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

Visual question answering (VQA) is challenging not only because the model hasto handle multi-modal information, but also because it is just so hard tocollect sufficient training examples -- there are too many questions one canask about an image. As a result, a VQA model trained solely on human-annotatedexamples could easily over-fit specific question styles or image contents thatare being asked, leaving the model largely ignorant about the sheer diversityof questions. Existing methods address this issue primarily by introducing anauxiliary task such as visual grounding, cycle consistency, or debiasing. Inthis paper, we take a drastically different approach. We found that many of the"unknowns" to the learned VQA model are indeed "known" in the datasetimplicitly. For instance, questions asking about the same object in differentimages are likely paraphrases; the number of detected or annotated objects inan image already provides the answer to the "how many" question, even if thequestion has not been annotated for that image. Building upon these insights,we present a simple data augmentation pipeline SimpleAug to turn this "known"knowledge into training examples for VQA. We show that these augmented examplescan notably improve the learned VQA models' performance, not only on the VQA-CPdataset with language prior shifts but also on the VQA v2 dataset without suchshifts. Our method further opens up the door to leverage weakly-labeled orunlabeled images in a principled way to enhance VQA models. Our code and dataare publicly available at https://github.com/heendung/simpleAUG.

Discovering the Unknown Knowns: Turning Implicit Knowledge in the Dataset into Explicit Training Examples for Visual Question Answering