Skip to main content

Table 1 Classification accuracy on the downstream tasks

From: Exploiting generative self-supervised learning for the assessment of biological images with lack of annotations

  GAN-DL Baseline[28] (%) ImageNet-collapsed (%) ImageNet-concatenated (%) ConvAE (%)
\(C^+\) versus \(C^-\) 91.4 95.81 65.31 79.61 64.50
HRCE versus VERO 100.0 100.0 85.52 100.0 99.80
\(C^+\) versus \(C^-\) (HRCE only) 92.44 99.28 84.09 90.24 68.41
\(C^+\) versus \(C^-\) (VERO only) 99.93 100 84.53 99.8 82.89