Out of Distribution Generalization via Interventional Style Transfer in Single-Cell Microscopy
Authors:
Wolfgang M. Pernice,
Michael Doron,
Alex Quach,
Aditya Pratapa,
Sultan Kenjeyev,
Nicholas De Veaux,
Michio Hirano,
Juan C. Caicedo
Abstract:
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 cau…
▽ More
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 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 and datasets.
△ Less
Submitted 15 June, 2023;
originally announced June 2023.
