Bridging the Gap: Differentially Private Equivariant Deep Learning for Medical Image Analysis
This work proposes to use steerable equivariant convolutional networks for medical image analysis with Differential Privacy, and demonstrates that ECNNs have superior performance with a narrowed privacy-utility gap, even at smaller model sizes.
A comparison of feature selection methods for the development of a prognostic radiogenomic biomarker in non-small cell lung cancer patients
This study aims at comparing methods for selecting optimal radiomic and gene expression features to develop a radiogenomic phenotype, that will be used to predict overall survival in non-small cell…
Equivariant Differentially Private Deep Learning
This work proposes to use more efﬁcient models with improved feature quality by in-troducing steerable equivariant convolutional networks for DP training and demonstrates that their models are able to outperform the current SOTA performance on CIFAR-10 by up to 9% across different ε -values while reducing the number of model parameters.