Algorithm 1: Pseudo code for hybridized ENN-GNDO-IPA algorithm.

Global Search Phase  Generalized normal distribution optimizer: Start  Input: Population size N, The Upper and Lower bounds (u,l). Current number of iteration is t and maximum number of iterations is (Max_iter). Initial population is developed randomly by the entries of real number with number of dimensions equal to unknown parameters in ENN structure. Weights = W = [${\alpha }_j$, ${\xi }_j$, ${\beta }_j$],   $j=1,2,3\dots ,n$.  Population: Generate population P of n candidates with the set of random weights drawn from a normal distribution as: P = [$C_1$, $C_2$, $C_3$, …, $C_m$]${}^t$, $\alpha$ = [${\alpha }_1$, ${\alpha }_2$, ${\alpha }_3$, …, ${\alpha }_n$], $\xi$ = [${\xi }_1$, ${\xi }_2$, ${\xi }_3$, …, ${\xi }_n$] and $\beta$ = [${\beta }_1$, ${\beta }_2$, ${\beta }_3$, …, ${\beta }_n$].  Output: Choose the current best solution i.e., $C_{GNG{O}_{Best}}$.  Initializations of GNDO: Initialize population P.  Fitness evaluation: Evaluate the fitness value using Equations (6), (12), and (24) for each individual of population C in P and achieve the so far best solution $x_{Best}$.             The iteration is updated as $t=t+1$.  Main Loop   while $(t\le (Max\_iter\left)\right)$  do     for if  $i=1:N$        p is randomly generated between 0 and 1.        if $p>0.5$        Exploitation Current best solution $x_{Best}$ is selected. $\eta$, d, m and M are evaluated using Equations (42)–(45) to execute the process.        else        Exploration The current best solution $x_{Best}$ is selected to perform exploration using Equations (46)–(48).        end if     end for          The iteration is updated as $t=t+1$.   end while  Termination: Terminate the algorithm: Predefined number of iterations is achieved. Fitness $ϵ\le {10}^{-20}$, TolFunc $ϵ\le {10}^{-25}$  Storage: Store global best weights $C_{GNG{O}_{Best}}$ and corresponding fitness values.  Generalized normal distribution Optimization: End  Local Search Phase  Interior Point Algorithm: Start  Inputs: IPT is incorporated for fine tuning of parameter by taking the best weights of GNDO as the start point.  Output: GNDO-IPA best weights i.e., $C_{GNGO-IPA}$  Initialization:    Start-Point as $C_{GNG{O}_{Best}}$ number of iterations, bound constraints.  Termination: Adaption process ends if any of the following conditions are met: Fitness $ϵ={10}^{-20}$, total iterations ≤ 2000 TolFun ≤${10}^{-25}$, TolX ≤${10}^{-25}$ TolCon ≤${10}^{-25}$, Max. Fun. Evaluations ≤ 200,000 while (satisfied the required termination) Fitness evaluation: Calculate fitness of each weight vector C. Fine-tuning: Use ‘fmincon’ and ‘optimset’ routines of the MATLAB optimization toolbox for IPA. Update parameters of C for each generation of IPA and calculate fitness ($\zeta$) of modified C.  Storage: Accumulate weights vector $C_{GNDO-IPA}$, fitness value, iterations, and function evaluations.  Interior point Algorithm: End  Data Generations: Repeat 100 times the procedure steps to generate a massive data set of the optimization variables of ENN to solve heat transfer and convection porous fin problems.