MOS: Towards Scaling Out-of-distribution Detection for Large Semantic Space

By Rui Huang, Yixuan Li. Computer Vision and Pattern Recognition, 2021

332 citations recorded by the source corpus.

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.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/f19092561296244e1dafe7d799e7906e96a63773

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

On the Importance of Gradients for Detecting Distributional Shifts in the Wild (2021, methodology): “4 Experiments In this section, we evaluate GradNorm on a large-scale OOD detection benchmark with ImageNet-1k as in-distribution dataset [16].”
ViM: Out-Of-Distribution with Virtual-logit Matching (2022, methodology): “For example, the Places and the SUN dataset selected by [18] have a large portion of images that are indistinguishable from ID samples.”
OpenOOD: Benchmarking Generalized Out-of-Distribution Detection (2022, methodology): “MOS [52] uses the prior of super category to perform hierarchical OOD detection.”
Out-of-distribution Detection with Deep Nearest Neighbors (2022, methodology): “On a challenging ImageNet OOD detection benchmark (Huang & Li, 2021), our KNN-based approach achieves superior performance under a similar inference speed as the baseline methods.”
ReAct: Out-of-distribution Detection With Rectified Activations (2021, methodology): “We evaluate on four test OOD datasets from (subsets of) Places365 [62], Textures [5], iNaturalist [16], and SUN [55] with non-overlapping categories w.r.t ImageNet.”
Generalized Out-of-Distribution Detection: A Survey (2021, unknown): “Under the redesigned label architecture, top-down classification strategy [167] and group softmax training [168] are demonstrated effective.”