Algorithm 2: Proposed SVM classifier
Input	Extracted feature map  $\mathit{x}=(\mathit{a}1,\mathit{a}2,.!,\mathit{a}\mathit{n})$  with annotations  $\mathit{a}=0,1$ . Test data Atest
Output	Distinguishing between diabetic retinopathy (DR) cases and normal diabetic photographic samples
Step 1	The SVM classifier parameters are specified to achieve optimization
Step 2	Multi level Classification of samples
Step 3	Conv2D was incorporated
Step 4	SVM-based classifier a. Our Algorithm 1 is utilized to complete the training process of SVM by extracting features represented as t = (a1, a2,…, an) b. The generation of the hyperplane through Equation (2)
Step 5	During the testing phase, the class label is assigned to the samples using a z-test based on the decision function given by the equation: Vout = (Weight_vector, Input_features) + bias_term