Overcoming language priors for visual question answering via loss rebalancing label and global context

R Cao, Z Li - Uncertainty in Artificial Intelligence, 2023 - proceedings.mlr.press
Despite the advances in Visual Question Answering (VQA), many VQA models currently
suffer from language priors (ie generating answers directly from questions without using …
被引用次数：2 相关文章

Overcoming Language Priors for Visual Question Answering via Loss Rebalancing Label and Global Context

R Cao, Z Li - The 39th Conference on Uncertainty in Artificial … - openreview.net
Despite the advances in Visual Question Answering (VQA), many VQA models currently
suffer from language priors (ie generating answers directly from questions without using …

Overcoming language priors for visual question answering via loss rebalancing label and global context

R Cao, Z Li - Proceedings of the Thirty-Ninth Conference on …, 2023 - dl.acm.org
Despite the advances in Visual Question Answering (VQA), many VQA models currently
suffer from language priors (ie generating answers directly from questions without using …

Overcoming Language Priors for Visual Question Answering via Loss Rebalancing Label and Global Context

R Cao, Z Li - auai.org
Our contributions:• We propose a novel model-agnostic generic framework LRLGC that
enables end-to-end training and can be easily integrated into various VQA models.• We …

Overcoming Language Priors for Visual Question Answering via Loss Rebalancing Label and Global Context

R Cao, Z Li - auai.org
Our contributions:• We propose a novel model-agnostic generic framework LRLGC that
enables end-to-end training and can be easily integrated into various VQA models.• We …