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

Models for Visual Question Answering (VQA) often rely on the spuriouscorrelations, i.e., the language priors, that appear in the biased samples oftraining set, which make them brittle against the out-of-distribution (OOD)test data. Recent methods have achieved promising progress in overcoming thisproblem by reducing the impact of biased samples on model training. However,these models reveal a trade-off that the improvements on OOD data severelysacrifice the performance on the in-distribution (ID) data (which is dominatedby the biased samples). Therefore, we propose a novel contrastive learningapproach, MMBS, for building robust VQA models by Making the Most of BiasedSamples. Specifically, we construct positive samples for contrastive learningby eliminating the information related to spurious correlation from theoriginal training samples and explore several strategies to use the constructedpositive samples for training. Instead of undermining the importance of biasedsamples in model training, our approach precisely exploits the biased samplesfor unbiased information that contributes to reasoning. The proposed method iscompatible with various VQA backbones. We validate our contributions byachieving competitive performance on the OOD dataset VQA-CP v2 while preservingrobust performance on the ID dataset VQA v2.

Towards Robust Visual Question Answering: Making the Most of Biased Samples via Contrastive Learning