Skip to main content

Table 9 4-Fold Cross Validation (%) of Different Frameworks on UCSB Dataset

From: Reverse active learning based atrous DenseNet for pathological image classification

  Single FV descriptor Combination of FV descriptors
  FV-SIFT FV-DBN FV-CNN S+D S+C D+C S+D+C
SVM 87.9 82.8 96.6 86.2 93.1 91.4 93.1
SDR+SVM [26] 89.7 89.7 98.3 91.4 94.8 96.6 94.8
ADN+DRAL (ours) 100
  1. Best accuracy is in Bold.