Algorithm 1: Pseudocode of DBAM-MCASVM-PCN Model.

Input: Input features $x_1, x_2, \dots , x_n$

Output: Classified result $y$

Residual = $\varnothing$; counter = $\varnothing$

layers = [MLP Feature Extraction, DBAM Module, MCASVM Classifier, Probabilistic Calibration Network]

1: While counter < len(layers) ∗ 2 + 1:

2:      If counter < len(layers):

3:         If counter == ∅:

4:          layer = MLP Feature Extraction

5:         ElseIf counter == 1:

6:           $F_{max}=Maxpool\left(F\right)$

7:           $F_{avg}=Avgpool\left(F\right)$

8:           ${MLP}_{F_{max}}={Share}_{MLP\left(F_{max}\right)}$

9:           ${MLP}_{F_{avg}}={Share}_{MLP\left(F_{avg}\right)}$

10:           $M_c=Sigmoid({MLP}_{F_{max}}\ast {MLP}_{F_{avg}})$

11:           $F^{\prime }=M_c\ast F$

12:           $F_{max}^{\prime }=MaxPool{(F}^{\prime })$

13:           $F_{avg}^{\prime }=AvgPool{(F}^{\prime })$

14:           $M_s=Sigmoid\left(Conv\right(F_{max}^{\prime }+F_{avg}^{\prime }\left)\right)$

15:           $F_{final}{=M}_s\ast M_c\ast F$

16:       ElseIf counter == 2:

17:          $d$ = []

18:          For each class k in {1, 2, …, K}:

19:              $d_k=$ $w_k$ ∗ $F_{final}$$+b_k$

20:              Append $d_k$ to $d$

21:       ElseIf counter == 3:

22:          p = Softmax(d)

23:          y = argmax(p)

24:           EndIf

25:           x = layer(x)

26:       Else:

27:            If counter == len(layers):

28:              layer = MCASVM Classifier

29:            ElseIf counter == len(layers) + 1:

30:              layer = Probabilistic Calibration Network

31:            EndIf

32:            x = layer(x)

33:          counter = counter + 1

34: y = Output Layer(x)

35: Return y