Skip to main content

Table 5 Accuracy and standard deviation scores obtained in the final iteration for clean and noisy datasets

From: A predicted-loss based active learning approach for robust cancer pathology image analysis in the workplace

Dataset

LC

Entropy

BALD

RS

Core-set

LL

Proposed

Clean

0.912 Ā± 0.014

0.909 Ā± 0.013

0.916 Ā± 0.007

0.907 Ā± 0.012

0.859 Ā± 0.017

0.896 Ā± 0.014

0.924 Ā± 0.005

Noisy

0.878 Ā± 0.027

0.877 Ā± 0.028

0.838 Ā± 0.094

0.886 Ā± 0.021

0.850 Ā± 0.018

0.863 Ā± 0.062

0.917 Ā± 0.004

  1. Abbreviations LC Least confidence, BALD Bayesian active learning by disagreement, RS Random sampling