a.	Parameter: [train, test, valid] = split(Dataset), a = argmax, m1 = model, t = train, v = valid, tt = test, b = batch, w = weights, em = ensemble-model, si = ith sample input of CXI dataset, X = class.
b.	Input: Dataset D = sum ${\sum }_{d=0}^n(t_{d,}{v}_d,{\mathit{tt}}_d)$
c.	Output: ensemble meta-classifier, c(si): predicted class index for ith image with PulDi-COVID. Model Training, Validation, testing:
1.	Action 1: Base model classification
2.	for c = [VGG16, ResNet50, VGG19, DenseNet201, MobileNetV2, NASNetMobile, ResNet152V2,
3.	DenseNet169]
4.	for j = 1:150 // epoch range from 1 to 150
5.	[t(j), v(j)] = partition(t, v)
6.	for k=(t/b, v/b) // sample/batch
7.	model(m1, c, j) = train(c(m1), a, t(j), v(j)) // training
8.	valid(m1, c, j) = valid((c(m1), a, j), v(j)) // validation
9.	end
10.	end
11.	tt(c(m1), a) = predict(max(test(c(m1), a, tt))) // prediction on test
12.	c(si) = tt(c(m1), a) // predicted class label with index
13.	end
14.	acquireWeights = model(w) //$w_{i,j}=\mathrm{m}\mathrm{a}\mathrm{x}({\bigcup }_{m=1}^M{\mathit{acc}}_{m,j})$
15.	Action 2: Create SSE ensemble-classifier(meta-model) and load feature vectors.
16.	for i = 1:8
17.	em = model(a). append(c(i))// input as base model structure to create meta-ensemble-learner
18.	em(w) = loadWeights(acquireWeights(c[1…8]))
19.	Action 3: Test ensemble classifier
20.	tt(em, a) = predict(max(test(em, a, tt))) // estimate the class probability
21.	$p_x\left(s_i\right)$ = tt(em, a)
22.	c(si) = $p_x\left(s_i\right)$// predicted specific class label with index
23.	end;