Most of the existing Out-Of-Distribution (OOD) detection algorithms depend on single input source: the feature, the logit, or the softmax probability. However, the immense diversity of the OOD examples makes such methods fragile. There are OOD samples that are easy to identify in the feature space while hard to distinguish in the logit space and vice versa. Motivated by this observation, we propose a novel OOD scoring method named Virtual-logit Matching (ViM), which combines the class-agnostic score from feature space and the In-Distribution (ID) class-dependent logits. Specifically, an additional logit representing the virtual OOD class is generated from the residual of the feature against the principal space, and then matched with the original logits by a constant scaling. The probability of this virtual logit after softmax is the indicator of OOD-ness. To facilitate the evaluation of large-scale OOD detection in academia, we create a new OOD dataset for ImageNet-1K, which is human-annotated and is 8.8x the size of existing datasets. We conducted extensive experiments, including CNNs and vision transformers, to demonstrate the effectiveness of the proposed ViM score. In particular, using the BiT-S model, our method gets an average AUROC 90.91% on four difficult OOD benchmarks, which is 4% ahead of the best baseline. Code and dataset are available at https://github.com/haoqiwang/vim.

本研究旨在解决已有的OutOfDistribution（OOD）检测算法在面对OOD的多样性时，容易出现一定程度的脆弱性问题。因此，我们提出了一种名为Virtual-logit Matching（ViM）的新型OOD评分方法，它能够结合来自特征空间的分类无关分数和In-Distribution（ID）类相关对数。我们为ImageNet-1K创建了一个新的OOD数据集，该数据集是人类注释的，大小为现有数据集的8.8倍。经过广泛的实验证明，我们的方法在四个难以检测的OOD基准测试中获得了优异的表现，特别是使用BiT-S模型，我们的方法在平均AUROC方面超越了最佳基线，达到了90.91％。

ViM：采用虚拟对数匹配进行离分布