arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2303.14531cs.CVcs.LG
收藏

SIO:合成分布内数据优势分布外检测

SIO: Synthetic In-Distribution Data Benefits Out-of-Distribution Detection

Duke University(杜克大学);Air Force Research Laboratory(空军研究实验室)

Jingyang Zhang, Nathan Inkawhich, Randolph Linderman, Ryan Luley, Yiran Chen, Hai Li

摘要:

构建可靠的分布外(OOD)检测器具有挑战性,通常需要在训练期间使用OOD数据。在这项工作中,我们开发了一种数据驱动的方法,该方法与现有的方法不同,并且是对现有方法的补充:我们不使用外部OOD数据,而是通过利用生成模型来充分利用内部分布(ID)训练集来生成额外的合成ID图像。然后,使用一种新的目标来训练分类器,该目标计算真实和合成ID样本的加权损失。我们的训练框架,这是所谓的SIO,作为一个“即插即用”的技术,旨在与现有的和未来的OOD检测算法,包括那些利用现有的OOD训练数据兼容。我们在CIFAR-10、CIFAR-100和ImageNet变体上的实验表明,SIO始终如一地提高了几乎所有最先进的(SOTA)OOD检测算法的性能。例如,在具有挑战性的CIFAR-10与在CIFAR-100检测问题上,SIO将现有的18种方法的平均OOD检测AUROC从86.25%提高到89.04%,达到了OpenOOD基准测试的92.94%的新SOTA。代码可在https://github.com/zjysteven/SIO上获得。

英文摘要:

Building up reliable Out-of-Distribution (OOD) detectors is challenging, often requiring the use of OOD data during training. In this work, we develop a data-driven approach which is distinct and complementary to existing works: Instead of using external OOD data, we fully exploit the internal in-distribution (ID) training set by utilizing generative models to produce additional synthetic ID images. The classifier is then trained using a novel objective that computes weighted loss on real and synthetic ID samples together. Our training framework, which is termed SIO, serves as a "plug-and-play" technique that is designed to be compatible with existing and future OOD detection algorithms, including the ones that leverage available OOD training data. Our experiments on CIFAR-10, CIFAR-100, and ImageNet variants demonstrate that SIO consistently improves the performance of nearly all state-of-the-art (SOTA) OOD detection algorithms. For instance, on the challenging CIFAR-10 v.s. CIFAR-100 detection problem, SIO improves the average OOD detection AUROC of 18 existing methods from 86.25\% to 89.04\% and achieves a new SOTA of 92.94\% according to the OpenOOD benchmark. Code is available at https://github.com/zjysteven/SIO.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑