Timezone: »

Boosting Out-of-distribution Detection with Typical Features
Yao Zhu · YueFeng Chen · Chuanlong Xie · Xiaodan Li · Rong Zhang · Hui Xue' · Xiang Tian · bolun zheng · Yaowu Chen

Wed Nov 30 09:00 AM -- 11:00 AM (PST) @ Hall J #133

Out-of-distribution (OOD) detection is a critical task for ensuring the reliability and safety of deep neural networks in real-world scenarios. Different from most previous OOD detection methods that focus on designing OOD scores or introducing diverse outlier examples to retrain the model, we delve into the obstacle factors in OOD detection from the perspective of typicality and regard the feature's high-probability region of the deep model as the feature's typical set. We propose to rectify the feature into its typical set and calculate the OOD score with the typical features to achieve reliable uncertainty estimation. The feature rectification can be conducted as a plug-and-play module with various OOD scores. We evaluate the superiority of our method on both the commonly used benchmark (CIFAR) and the more challenging high-resolution benchmark with large label space (ImageNet). Notably, our approach outperforms state-of-the-art methods by up to 5.11% in the average FPR95 on the ImageNet benchmark.

Author Information

Yao Zhu (Zhejiang University)
YueFeng Chen (Alibaba Group)
Chuanlong Xie (Beijing Normal University)
Xiaodan Li (University of Science and Technology of China)
Rong Zhang (Huazhong University of Science and Technology)
Hui Xue' (Zhejiang University, Tsinghua University)
Xiang Tian (Zhejiang University)
bolun zheng (Hangzhou Dianzi University)
Yaowu Chen

More from the Same Authors