Understanding Rare Spurious Correlations in Neural Networks
This addresses privacy and robustness issues in neural networks for AI practitioners, though it is incremental as it builds on prior work on spurious correlations.
The paper investigates neural networks' sensitivity to rare spurious correlations, finding that only a few training examples can cause networks to learn these correlations, which negatively impact accuracy and privacy.
Neural networks are known to use spurious correlations such as background information for classification. While prior work has looked at spurious correlations that are widespread in the training data, in this work, we investigate how sensitive neural networks are to rare spurious correlations, which may be harder to detect and correct, and may lead to privacy leaks. We introduce spurious patterns correlated with a fixed class to a few training examples and find that it takes only a handful of such examples for the network to learn the correlation. Furthermore, these rare spurious correlations also impact accuracy and privacy. We empirically and theoretically analyze different factors involved in rare spurious correlations and propose mitigation methods accordingly. Specifically, we observe that $\ell_2$ regularization and adding Gaussian noise to inputs can reduce the undesirable effects. Code available at https://github.com/yangarbiter/rare-spurious-correlation.