Ensemble based in transfer learning for cytological classification in pleural fluid.

However, on the larger external Kaggle dataset ( n = 693), the differences between the ResNet + DenseNet ensemble and the individual models were highly significant: compared to ResNet50V2 ( Z = 4.816, p < 0.001) and InceptionV3 ( Z = 8.042, p < 0.001).

also in 132,142 other papers

The Z -test for two proportions revealed that the accuracy differences between ensemble and individual models on the local dataset did not reach statistical significance ( p > 0.05 for all comparisons), which is consistent with the limited statistical power afforded by a test set of 388 images when baseline accuracies exceed 94%.

also in 111,027 other papers