Table 3 Overview of feature selection and classifier methods used
From: Predictive performance of radiomic models based on features extracted from pretrained deep networks
Method | Hyperparameter | |
|---|---|---|
Feature selection | ANOVA | – |
Bhattacharyya distance | – | |
Extra trees | Trees = 100 | |
LASSO | C = 1 | |
Random Forest | Trees = 100 | |
t-Score | – | |
Classifier | Logistic regression | C in 2^{− 6, − 4, − 2, 0, 2, 4, 6} |
Naive Bayes | – | |
Neural network | Three layers with 4, 16 or 64 neurons each | |
Random forest | Number of estimators 50, 125 or 250 | |
Support vector machines | C in 2^{− 6, − 4, − 2, 0, 2, 4, 6}, gamma was determined automatically |