Robustness to Spurious Correlations Improves Semantic Out-of-Distribution Detection
This addresses a critical issue for real-world deployment of OOD detection systems, where anomalies often occur in similar settings as normal inputs, though it appears incremental as it builds on existing OOD generalization techniques.
The paper tackles the problem of detecting out-of-distribution (OOD) inputs that share nuisance features (e.g., backgrounds) with in-distribution data, proposing nuisance-aware OOD detection to address failures in existing methods. The result shows that this approach substantially improves performance over original methods, even outperforming domain generalization algorithms.
Methods which utilize the outputs or feature representations of predictive models have emerged as promising approaches for out-of-distribution (OOD) detection of image inputs. However, these methods struggle to detect OOD inputs that share nuisance values (e.g. background) with in-distribution inputs. The detection of shared-nuisance out-of-distribution (SN-OOD) inputs is particularly relevant in real-world applications, as anomalies and in-distribution inputs tend to be captured in the same settings during deployment. In this work, we provide a possible explanation for SN-OOD detection failures and propose nuisance-aware OOD detection to address them. Nuisance-aware OOD detection substitutes a classifier trained via empirical risk minimization and cross-entropy loss with one that 1. is trained under a distribution where the nuisance-label relationship is broken and 2. yields representations that are independent of the nuisance under this distribution, both marginally and conditioned on the label. We can train a classifier to achieve these objectives using Nuisance-Randomized Distillation (NuRD), an algorithm developed for OOD generalization under spurious correlations. Output- and feature-based nuisance-aware OOD detection perform substantially better than their original counterparts, succeeding even when detection based on domain generalization algorithms fails to improve performance.