Abstract:Out-of-distribution (OOD) detection is indispensable for deploying reliable machine learning systems in real-world scenarios. Recent works, using auxiliary outliers in training, have shown good potential. However, they seldom concern the intrinsic correlations between in-distribution (ID) and OOD data. In this work, we discover an obvious correlation that OOD data usually possesses significant ID attributes. These attributes should be factored into the training process, rather than blindly suppressed as in previous approaches. Based on this insight, we propose a structured multi-view-based out-of-distribution detection learning framework, MVOL, which facilitates rational handling of the intrinsic in-distribution attributes in outliers. We provide theoretical insights on the effectiveness of MVOL for OOD detection. Extensive experiments demonstrate the superiority of our framework to others. MVOL effectively handles both auxiliary OOD datasets and even wild datasets with indistribution data as noise.
This project is for the paper: Mining In-distribution Attributes in Outliers for Out-of-distribution Detection published at AAAI 2025. Some parts of the codebase are adapted from Outlier Exposure, and NTOM.
The following packages are required to be installed:
Our experiments are conducted with Python 3.8 on NVIDIA 2080 GPUs.
- In-distribution training set:
- CIFAR: included in PyTorch.
-
Auxiliary outlier training set:
- 300K Random Images:provided by Outlier Exposure.
- [Mixed CIFAR100]:auxiliary datasets used in wild settings included in this repository(datasets/wild_datasets).
We provide links and instructions to download each dataset:
- SVHN: download it and place it in the folder of
datasets/test_ood_datasets/svhn. Then runpython select_svhn_data.pyto generate test subset. - Textures: download it and place it in the folder of
datasets/test_ood_datasets/dtd. - Places365: download it and place it in the folder of
datasets/test_ood_datasets/places365/test_subset. We randomly sample 10,000 images from the original test dataset. - LSUN: download it and place it in the folder of
datasets/test_ood_datasets/LSUN. - LSUN-resize: download it and place it in the folder of
datasets/test_ood_datasets/LSUN_resize. - iSUN: download it and place it in the folder of
datasets/test_ood_datasets/iSUN.
Each experiment is conducted in both Single Model Setting and Ensemble Distillation Model Setting with 5 independently runs. Reported results in paper are average of them. Note Single Model Setting is usually adopted in literature.
sh scripts/run_cifar10.sh
sh scripts/run_cifar100.sh
sh scripts/run_wild.sh
If you find this work helps your research, please consider citing our paper:
@misc{lei2024miningindistributionattributesoutliers,
title={Mining In-distribution Attributes in Outliers for Out-of-distribution Detection},
author={Yutian Lei and Luping Ji and Pei Liu},
year={2024},
eprint={2412.11466},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2412.11466},
}