Out of Distribution Generalization via Interventional Style Transfer in Single-Cell Microscopy
Out of Distribution Generalization via Interventional Style Transfer in Single-Cell Microscopy
复制标题
DOI：
10.1109/cvprw59228.2023.00455
复制
发表时间：
2023-06
期刊：
影响因子：
--
通讯作者：
Wolfgang M. Pernice;Michael Doron;A. Quach;Aditya Pratapa;Sultan Kenjeyev;N. Veaux;Michio Hirano;Juan C. Caicedo
中科院分区：
文献类型：
--
作者：
Wolfgang M. Pernice;Michael Doron;A. Quach;Aditya Pratapa;Sultan Kenjeyev;N. Veaux;Michio Hirano;Juan C. Caicedo
计算机视觉系统的现实世界部署，包括在生物医学研究的发现过程中，需要对上下文干扰不变并推广到新数据的因果表示。利用两个新颖的单细胞荧光显微镜数据集的内部复制结构，我们提出了普遍适用的测试来评估模型在日益具有挑战性的 OOD 泛化水平上学习因果表示的程度。我们表明，尽管通过其他既定指标评估的表现看似强劲，但旨在防止混淆的幼稚基线和当代基线在这些测试中都崩溃为随机的。我们引入了一种新方法，即介入风格迁移（IST），它通过生成介入训练分布来显着提高 OOD 泛化能力，其中生物学原因和滋扰之间的虚假相关性得到了缓解。我们发布代码 1 和数据集 2。
Real-world deployment of computer vision systems, including in the discovery processes of biomedical research, requires causal representations that are invariant to contextual nuisances and generalize to new data. Leveraging the internal replicate structure of two novel single-cell fluorescent microscopy datasets, we propose generally applicable tests to assess the extent to which models learn causal representations across increasingly challenging levels of OOD-generalization. We show that despite seemingly strong performance as assessed by other established metrics, both naive and contemporary baselines designed to ward against confounding, collapse to random on these tests. We introduce a new method, Interventional Style Transfer (IST), that substantially improves OOD generalization by generating interventional training distributions in which spurious correlations between biological causes and nuisances are mitigated. We publish our code 1 and datasets 2.