Skip to main content

Table 2 The relevant parameters of conventional methods

From: Automatic diagnosis of imbalanced ophthalmic images using a cost-sensitive deep convolutional neural network

Classifiers Features extraction methods [27, 29] Data-level methods [18, 19, 22]
Random forest (RF): number of trees = 300
Support vector machine (SVM): linear kernel function
Color and texture features (COTX): gray tone spatial dependence matrices: d = 1; gray gradient co-occurrence matrices: Lg = 10 SMOTE and BorSMOTE: nearest neighbors k = 5, the ratio of positive and negative samples r = 1
Under-sampling: the ratio of positive and negative samples r = 1
Local binary pattern (LBP): P = 9
Wavelet transformation (WAVE): two level wavelet transformation, Haar wavelet
Scale-invariant feature transform (SIFT)