Causal attention for unbiased visual recognition

• المؤلفون الرئيسيون: WANG, Tan, ZHOU, Chang, SUN, Qianru, ZHANG, Hanwang
• التنسيق: text
• اللغة: English
• منشور في: Institutional Knowledge at Singapore Management University 2021
• الموضوعات: Graphics and Human Computer Interfaces
• الوصول للمادة أونلاين: https://ink.library.smu.edu.sg/sis_research/6228; https://ink.library.smu.edu.sg/context/sis_research/article/7231/viewcontent/Wang_Causal_Attention_for_Unbiased_Visual_Recognition_ICCV_2021_paper.pdf

Attention module does not always help deep models learn causal features that are robust in any confounding context, e.g., a foreground object feature is invariant to different backgrounds. This is because the confounders trick the attention to capture spurious correlations that benefit the predictio...