Overcoming Language Priors via Shuffling Language Bias for Robust Visual Question Answering

Overcoming Language Priors via Shuffling Language Bias for Robust Visual Question Answering

Recent research has revealed the notorious language prior problem in visual question answering (VQA) tasks based on visual-textual interaction, which indicates that well-developed VQA models rely on learning shortcuts from questions without fully considering visual evidence.To tackle this problem, m...

Full description

Bibliographic Details
Main Authors: J. Zhao, Z. Yu, X. Zhang, Y. Yang
Format: Article
Language:English
Published: IEEE 2023-01-01
Series:IEEE Access
Subjects:
Visual question answering
language prior
data balance
data augmentation
Online Access:https://ieeexplore.ieee.org/document/10214540/

Internet

https://ieeexplore.ieee.org/document/10214540/

Similar Items