Mitigating Biases with Diverse Ensembles and Diffusion Models
arxiv(2023)
摘要
Spurious correlations in the data, where multiple cues are predictive of the
target labels, often lead to a phenomenon known as shortcut bias, where a model
relies on erroneous, easy-to-learn cues while ignoring reliable ones. In this
work, we propose an ensemble diversification framework exploiting Diffusion
Probabilistic Models (DPMs) for shortcut bias mitigation. We show that at
particular training intervals, DPMs can generate images with novel feature
combinations, even when trained on samples displaying correlated input
features. We leverage this crucial property to generate synthetic
counterfactuals to increase model diversity via ensemble disagreement. We show
that DPM-guided diversification is sufficient to remove dependence on primary
shortcut cues, without a need for additional supervised signals. We further
empirically quantify its efficacy on several diversification objectives, and
finally show improved generalization and diversification performance on par
with prior work that relies on auxiliary data collection.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要