Algorithm 4: Extreme Gradient Decision-Boosting Machine (xgBoost) algorithm

Input: Extracted feature data $x=(x_1,x_2,\dots ,x_n)$ with labels y and test data $x_{test}$

Output: Class labels (glaucoma (GA), diabetic retinopathy (DR), cataract (CT), and normal (NL)).

Process:

Step 1. First, initialize tree as a constant: $y_i^t=f_0=0$ for optimization and parameter defined for the classifier.

Step 2. Construct XgBoost tree by minimizing the lost function using Equation (2).

The XgBoost() classifier is perfectly trained on the feature samples $x=(x_1,x_2,\dots ,x_n)$.

Step 3. Repeat Step 2 until the model reaches the stop condition.

Step 4. Test samples $y_i^t$ are assigned the class label using the decision function of Equation (2).