Mixture Outlier Exposure: Towards Out-of-Distribution Detection in Fine-grained Environments
Mixture Outlier Exposure: Towards Out-of-Distribution Detection in Fine-grained Environments
复制标题
DOI：
10.1109/wacv56688.2023.00549
复制
发表时间：
2021-06
期刊：
影响因子：
--
通讯作者：
Jingyang Zhang;Nathan Inkawhich;Randolph Linderman;Yiran Chen;H. Li
中科院分区：
文献类型：
--
作者：
Jingyang Zhang;Nathan Inkawhich;Randolph Linderman;Yiran Chen;H. Li
部署基于DNN的识别系统的许多现实场景具有固有的细粒度属性（例如，鸟类识别、医学图像分类）。除了实现可靠的准确性，这些模型的一个关键子任务是检测出的分布（OOD）的输入。考虑到部署环境的性质，人们可能期望这样的OOD输入也是细粒度的w.r.t.已知的类（例如，一种新的鸟类），因此极难识别。不幸的是，细粒度场景中的OOD检测仍然在很大程度上未被探索。在这项工作中，我们的目标是填补这一空白，首先仔细构建四个大规模的细粒度的测试环境，在现有的方法被证明有困难。特别是，我们发现，即使在训练过程中显式地结合一组不同的辅助离群数据，也不能在细粒度OOD样本所在的广泛区域提供足够的覆盖范围。然后，我们提出了混合离群值暴露（MixOE），它将ID数据和训练离群值混合在一起，以扩大不同OOD粒度的覆盖范围，并训练模型，使预测置信度随着输入从ID过渡到OOD而线性衰减。大量的实验和分析表明，MixOE的有效性，建立在细粒度环境中的OOD检测器。该代码可在https://github.com/zjysteven/MixOE上获得。
Many real-world scenarios in which DNN-based recognition systems are deployed have inherently fine-grained attributes (e.g., bird-species recognition, medical image classification). In addition to achieving reliable accuracy, a critical subtask for these models is to detect Out-of-distribution (OOD) inputs. Given the nature of the deployment environment, one may expect such OOD inputs to also be fine-grained w.r.t. the known classes (e.g., a novel bird species), which are thus extremely difficult to identify. Unfortunately, OOD detection in fine-grained scenarios remains largely underexplored. In this work, we aim to fill this gap by first carefully constructing four large-scale fine-grained test environments, in which existing methods are shown to have difficulties. Particularly, we find that even explicitly incorporating a diverse set of auxiliary outlier data during training does not provide sufficient coverage over the broad region where fine-grained OOD samples locate. We then propose Mixture Outlier Exposure (MixOE), which mixes ID data and training outliers to expand the coverage of different OOD granularities, and trains the model such that the prediction confidence linearly decays as the input transitions from ID to OOD. Extensive experiments and analyses demonstrate the effectiveness of MixOE for building up OOD detector in finegrained environments. The code is available at https://github.com/zjysteven/MixOE.