Timezone: »
We show that for a wide class of harmonization/domain-invariance schemes several undesirable properties are unavoidable. If a predictive machine is made invariant to a set of domains, the accuracy of the output predictions (as measured by mutual information) is limited by the domain with the least amount of information to begin with. If a real label value is highly informative about the source domain, it cannot be accurately predicted by an invariant predictor. These results are simple and intuitive, but we believe that it is beneficial to state them for medical imaging harmonization.
Author Information
Daniel Moyer (MIT)
More from the Same Authors
-
2021 : Bayesian Image Reconstruction using Deep Generative Models »
Razvan Marinescu · Daniel Moyer · Polina Golland -
2022 Workshop: Medical Imaging meets NeurIPS »
DOU QI · Konstantinos Kamnitsas · Yuankai Huo · Xiaoxiao Li · Daniel Moyer · Danielle Pace · Jonas Teuwen · Islem Rekik -
2021 : Bayesian Image Reconstruction using Deep Generative Models »
Razvan Marinescu · Daniel Moyer · Polina Golland -
2019 Poster: Fast structure learning with modular regularization »
Greg Ver Steeg · Hrayr Harutyunyan · Daniel Moyer · Aram Galstyan -
2019 Spotlight: Fast structure learning with modular regularization »
Greg Ver Steeg · Hrayr Harutyunyan · Daniel Moyer · Aram Galstyan -
2019 Poster: Exact Rate-Distortion in Autoencoders via Echo Noise »
Rob Brekelmans · Daniel Moyer · Aram Galstyan · Greg Ver Steeg -
2018 Poster: Invariant Representations without Adversarial Training »
Daniel Moyer · Shuyang Gao · Rob Brekelmans · Aram Galstyan · Greg Ver Steeg