Fig. 2: Perturbation accuracy of the test sets using PCA-LDA (green), SVM (red), and CNN (blue) via three separate perturbation methods. | BJC Reports

Fig. 2: Perturbation accuracy of the test sets using PCA-LDA (green), SVM (red), and CNN (blue) via three separate perturbation methods.

From: System transferability of Raman-based oesophageal tissue classification using modern machine learning to support multi-centre clinical diagnostics

Fig. 2

a Effect of applying wavenumber shift perturbation (cm−1). b Effect of applying the objective perturbation, given as a percentage of the total estimated objective contribution that was added. c Effect of applying the response perturbation (%), given as the percentage change induced across the data. Mean accuracies are reported +/− 1 SD. Black dotted lines correspond to the maximum response perturbation observed between the three instruments (detailed in SI5).

Back to article page