Rethinking Out-of-Distribution Detection on Imbalanced Data Distribution
This addresses a critical issue for deploying reliable neural networks in real-world scenarios where data imbalance is common, offering a novel solution to enhance OOD detection under such conditions.
The paper tackles the problem of out-of-distribution (OOD) detection in neural networks when in-distribution data is imbalanced, identifying that existing methods misclassify tail class ID samples as OOD and OOD samples as head class ID. It introduces a theoretical framework and a training-time regularization technique that improves OOD detection performance on benchmarks like CIFAR10-LT, CIFAR100-LT, and ImageNet-LT against state-of-the-art approaches.
Detecting and rejecting unknown out-of-distribution (OOD) samples is critical for deployed neural networks to void unreliable predictions. In real-world scenarios, however, the efficacy of existing OOD detection methods is often impeded by the inherent imbalance of in-distribution (ID) data, which causes significant performance decline. Through statistical observations, we have identified two common challenges faced by different OOD detectors: misidentifying tail class ID samples as OOD, while erroneously predicting OOD samples as head class from ID. To explain this phenomenon, we introduce a generalized statistical framework, termed ImOOD, to formulate the OOD detection problem on imbalanced data distribution. Consequently, the theoretical analysis reveals that there exists a class-aware bias item between balanced and imbalanced OOD detection, which contributes to the performance gap. Building upon this finding, we present a unified training-time regularization technique to mitigate the bias and boost imbalanced OOD detectors across architecture designs. Our theoretically grounded method translates into consistent improvements on the representative CIFAR10-LT, CIFAR100-LT, and ImageNet-LT benchmarks against several state-of-the-art OOD detection approaches. Code is available at https://github.com/alibaba/imood.