%0 Conference Paper %1 Sun2022ICMLDeepNN %A Sun, Yiyou %A Ming, Yifei %A Zhu, Xiaojin %A Li, Yixuan %B Proceedings of the 39th International Conference on Machine Learning %D 2022 %E Chaudhuri, Kamalika %E Jegelka, Stefanie %E Song, Le %E Szepesvari, Csaba %E Niu, Gang %E Sabato, Sivan %I PMLR %K OOD_detection %P 20827--20840 %T Out-of-Distribution Detection with Deep Nearest Neighbors %U https://proceedings.mlr.press/v162/sun22d.html %V 162 %X Out-of-distribution (OOD) detection is a critical task for deploying machine learning models in the open world. Distance-based methods have demonstrated promise, where testing samples are detected as OOD if they are relatively far away from in-distribution (ID) data. However, prior methods impose a strong distributional assumption of the underlying feature space, which may not always hold. In this paper, we explore the efficacy of non-parametric nearest-neighbor distance for OOD detection, which has been largely overlooked in the literature. Unlike prior works, our method does not impose any distributional assumption, hence providing stronger flexibility and generality. We demonstrate the effectiveness of nearest-neighbor-based OOD detection on several benchmarks and establish superior performance. Under the same model trained on ImageNet-1k, our method substantially reduces the false positive rate (FPR@TPR95) by 24.77% compared to a strong baseline SSD+, which uses a parametric approach Mahalanobis distance in detection. Code is available: https://github.com/deeplearning-wisc/knn-ood.

@inproceedings{Sun2022ICMLDeepNN, abstract = {Out-of-distribution (OOD) detection is a critical task for deploying machine learning models in the open world. Distance-based methods have demonstrated promise, where testing samples are detected as OOD if they are relatively far away from in-distribution (ID) data. However, prior methods impose a strong distributional assumption of the underlying feature space, which may not always hold. In this paper, we explore the efficacy of non-parametric nearest-neighbor distance for OOD detection, which has been largely overlooked in the literature. Unlike prior works, our method does not impose any distributional assumption, hence providing stronger flexibility and generality. We demonstrate the effectiveness of nearest-neighbor-based OOD detection on several benchmarks and establish superior performance. Under the same model trained on ImageNet-1k, our method substantially reduces the false positive rate (FPR@TPR95) by 24.77% compared to a strong baseline SSD+, which uses a parametric approach Mahalanobis distance in detection. Code is available: https://github.com/deeplearning-wisc/knn-ood.}, added-at = {2023-07-10T04:51:22.000+0200}, author = {Sun, Yiyou and Ming, Yifei and Zhu, Xiaojin and Li, Yixuan}, biburl = {https://www.bibsonomy.org/bibtex/2280fa11816a397b8e5f10f53c4798d3b/andolab}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, interhash = {89254e84faf4ade5202c512733769cb3}, intrahash = {280fa11816a397b8e5f10f53c4798d3b}, keywords = {OOD_detection}, month = {17--23 Jul}, pages = {20827--20840}, pdf = {https://proceedings.mlr.press/v162/sun22d/sun22d.pdf}, publisher = {PMLR}, series = {Proceedings of Machine Learning Research}, timestamp = {2023-07-10T04:51:22.000+0200}, title = {Out-of-Distribution Detection with Deep Nearest Neighbors}, url = {https://proceedings.mlr.press/v162/sun22d.html}, volume = 162, year = 2022 }