Table 2 Accuracy and training runtime for LR like models with clipped ReLu activation function, and trained in a privacy-preserving manner using the protocols proposed in this paper
From: High performance logistic regression for privacy-preserving genome analysis
| Â | # features | # pos. samples | # neg. samples | # of iterations | Fivefold CV accuracy (%) | Avg. runtime (s) |
|---|---|---|---|---|---|---|
BC-TCGA | 17,814 | 422 | 48 | 10 | 99.58 | 2.52 |
GSE2034 | 12,634 | 142 | 83 | 223 | 64.82 | 26.90 |