MOS: Towards Scaling Out-of-distribution Detection for Large Semantic Space
It addresses the problem of safely deploying machine learning models in real-world applications by scaling OOD detection to large semantic spaces, representing an incremental advance over existing small-dataset solutions.
The paper tackles out-of-distribution detection for large-scale image classification by proposing a group-based framework with a novel scoring function, MOS, which reduces average FPR95 by 14.33% and achieves a 6x speedup in inference compared to previous methods.
Detecting out-of-distribution (OOD) inputs is a central challenge for safely deploying machine learning models in the real world. Existing solutions are mainly driven by small datasets, with low resolution and very few class labels (e.g., CIFAR). As a result, OOD detection for large-scale image classification tasks remains largely unexplored. In this paper, we bridge this critical gap by proposing a group-based OOD detection framework, along with a novel OOD scoring function termed MOS. Our key idea is to decompose the large semantic space into smaller groups with similar concepts, which allows simplifying the decision boundaries between in- vs. out-of-distribution data for effective OOD detection. Our method scales substantially better for high-dimensional class space than previous approaches. We evaluate models trained on ImageNet against four carefully curated OOD datasets, spanning diverse semantics. MOS establishes state-of-the-art performance, reducing the average FPR95 by 14.33% while achieving 6x speedup in inference compared to the previous best method.