Table - PMC

Skip to main content

View full-text article in PMC

. 2022 Sep 29;10(10):2438. doi: 10.3390/biomedicines10102438

Algorithm 1: Training FAM network

FN_T = Training (F, l)//FN_T-Trained FAM Network
//F: Feature Vector, l: Label → Class of the feature vector.
Initialize $ρ$ , NE = 0// $ρ$ = Vigilance Parameter, NE: Count training epochs
While NE < number of training epochs
I_i = F = (a₁,a₂,…,a_d);//d: dimension of feature vector
AI₁ = (a₁, a₂,…,a_d,1-a₁,1-a₂,…,1-a_d)
if AI = first input of label l
W = AI
W←l
else for (all j) compute T_j(AI) = $|A I \land W_{j}| / α + |W_{j}|$
J = argmax (T_j(AI); J: Winner node
if $|A I \land W_{j}| / |A I| \geq ρ$ ; vigilance test
if l = J: $W_{k}^{n e w} = W_{k}^{o l d} + β (A I \land W_{k}^{o l d})$
else: $ρ = M F_{k} (A I) + \in$
While (more winner nodes are available)
While (more training patterns are available)
End