Generative Interventions for Causal Learning
We introduce a framework for learning robust visual representations that generalize to new viewpoints, backgrounds, and scene contexts. Discriminative models often learn naturally occurring spurious correlations, which cause them to fail on images outside of the training distribution. In this paper, we show that we can steer generative models to manufacture interventions on features caused by confounding factors. Experiments, visualizations, and theoretical results show this method learns robust representations more consistent with the underlying causal relationships. Our approach improves performance on multiple datasets demanding out-of-distribution generalization, and we demonstrate state-of-the-art performance generalizing from ImageNet to ObjectNet dataset.
PDF Abstract CVPR 2021 PDF CVPR 2021 AbstractCode
ImageNet
ImageNet-C
ObjectNet
Stylized ImageNet
Results from the Paper
Ranked #44 on
Image Classification
on ObjectNet
(using extra training data)
