Volume,3,Issue-5,May2025 International Journal of Modern Science and Research Technology

ISSN NO-2584-2706

Enhancing Moss Growth Optimization with Genetic

Algorithm for Improved Performance
Kaushal Thakre; Purva Landge; Sandhya Dahake
Department of Master in Computer Application
G H Raisoni College of Engineering & Management Nagpur,
Maharashtra, India

Abstract: This study proposes MGO-GA, a The solutions generated by MGO are refined
hybrid optimization approach that combines with GA’s global search capability, reducing
Genetic Algorithm (GA) with Moss Growth the risk of getting trapped in local optima [6].
Optimization (MGO). In complex search Experimental results on benchmark functions
environments, MGO often suffers from showed that MGO hybridized with GA
premature convergence and stagnation, despite achieved better accuracy, stability and
its great exploration and exploitation convergence rate as compared to standard
capabilities. The suggested approach improves MGO algorithm, making it more reliable and
population variety and speeds up convergence efficient optimization approach.
by introducing genetic operators crossover and
mutation into the MGO process to get around 2. Literature Review
these problems. By using GA's global search 2.1 Selection of Moss Growth Optimization
ability, the hybrid MGO-GA enhances the (MGO):
results generated by MGO and reduces the Moss growth optimization was a newly
possibility of being trapped in local optima. developed nature inspired meta heuristic
algorithm that mimics the adaptive growth
Keywords: Hybridization, Benchmark waiver of moss it effectively balances the
functions, Convergence, Exploration and exploration (searching new areas in the
Exploitation. solution space) and exploitation (refining the
best solutions found so far) [8]. MGO was
1. Introduction chosen for this study due to its strong
Moss Growth Optimization (MGO) is a optimization capabilities and efficient
nature-based metaheuristic algorithm which exploration-exploitation balance. Its promising
known for its strong exploration and results in various benchmark functions further
exploitation abilities [1]. However, like many justified its selection [6].
other optimization techniques, MGO suffered
from premature convergence and stagnation, 2.2 Justification for Hybridization with
restricting its effectiveness in complex search Genetic Algorithm (GA):
spaces [2]. These challenges raised due to a GA was selected for hybridization with MGO
lack of variations in candidate solutions, due to its strong global search capability and
reducing the MGO’s ability to escape local genetic operators, which enhance solution
optima [4]. To overcome these limitations, a diversity and prevent premature convergence
hybrid MGO-GA (Moss Growth Optimization [9]. By incorporating crossover and mutation,
with Genetic Algorithm) approach was GA improves exploration, helping MGO
proposed, integrating genetic operators such as escape local optima and achieve better
crossover and mutation to enhance diversity optimization performance [10].
and improve convergence speed [3].

IJMSRT25MAY105 www.ijmsrt.com 592

DOI: https://doi.org/10.5281/zenodo.15571838
Volume,3,Issue-5,May2025 International Journal of Modern Science and Research Technology
ISSN NO-2584-2706

2.3 Classification of Optimization Algorithms: 2.5 Pseudo Code:

Algorithms are generally classified into natural, The Hybrid MGO-GA algorithm combines
genetic, human-based, and revolutionary. Moss Growth Optimization (MGO) with a
Natural algorithms are inspired by biological Genetic Algorithm (GA) to enhance solution
processes [1]. Genetic algorithms are based on accuracy and convergence speed. MGO
the principle of natural selection [3]. Human- explores the search space, and GA refines the
based algorithms leverage human intelligence best solution using selection, crossover, and
[4], and Revolutionary algorithms introduce mutation. The convergence curves of both
innovative computational methods [5]. algorithms are merged for performance
analysis. This hybrid approach improves
optimization efficiency by leveraging MGO’s
exploration and GA’s genetic diversity.

2.5.1 Code:
Function Hybrid_MGO_GA(N, MaxFEs, lb, ub,
dim, fobj):
Initialize random seed for reproducibility
// Step 1: Run MGO Algorithm
Print "Running MGO..."
Call MGO(N, MaxFEs, lb, ub, dim, fobj)
Store results in Best_pos_MGO,
Best_score_MGO, Convergence_curve_MGO
Print Best_score_MGO
Fig 1. Nature-inspired algorithm classification
// Step 2: Run GA for Fine-Tuning MGO’s Best
2.4 Algorithms and Authors: Solution
The Table1 represents various previously Print "Refining with GA..."
developed nature-inspired algorithms. These Set GA_Iterations to 100
algorithms are used in solving complex Call GA(Best_pos_MGO, lb, ub, dim, fobj,
optimization problems by mimicking natural GA_Iterations)
phenomena, biological behaviors, or physical Store results in Best_pos, Best_score,
principles. Convergence_curve_GA
// Step 3: Merge Convergence Data
Table 1: Algorithm, Authors & Year of publishing Find minimum length of
Sr. Algorithm Name Author Name Year Convergence_curve_MGO and
No. Convergence_curve_GA
1. Ant Colony Marco Dorigo 1992 Resize both convergence curves to this
Optimization (ACO) minimum length
2. Grey Wolf Optimizer Seyedali 2014 Convert both to column vectors
(GWO) Mirjalili et al
3. Moss Growth Ali Asghar 2024
Concatenate the two curves side-by-side into
Optimization (MGO) Heidari et al Convergence_curve
4. Genetic Algorithm Jhom Holland 1975 Print Best_score
(GA) Function GA(initial_pos, lb, ub, dim, fobj,
5. Differential Evolution Rainer Storn 1997 maxGen):
et al
Set population size to 20
6 Particle Swarm James 1995
Optimization Kennedy et al Set mutation rate to 0.1
7. Sine Cosine Algorithm Seyedali 2016 Set crossover rate to 0.8
Mirjalili Initialize population around initial_pos
Apply boundary constraints to population

IJMSRT25MAY105 www.ijmsrt.com 593

DOI: https://doi.org/10.5281/zenodo.15571838
Volume,3,Issue-5,May2025 International Journal of Modern Science and Research Technology
ISSN NO-2584-2706

2.6 Functions and Equations

Evaluate fitness of population Benchmark functions are standardized
Store best score and best position mathematical models that are used to evaluate
Initialize Convergence_curve optimization algorithms. Various search space
For each generation in maxGen: complexities, including unimodal, multimodal,
Perform tournament selection to choose parents and non-convex landscapes, are represented to
Perform crossover to generate offspring assess an algorithm’s efficiency in exploration
Perform mutation on offspring and convergence. The ability of an algorithm to
Evaluate fitness of offspring escape local optima and find the global
Replace worst individuals in population with optimum in diverse optimization scenarios is
better offspring tested using these functions. Smooth, rugged,
Update best score and best position if a better and deceptive landscapes are included to ensure
solution is found a comprehensive performance evaluation.
Store best score in Convergence_curve
Function tournamentSelection(population, Table 2: Standard UM Benchmark Functions
fitness, popSize):
Initialize parents array
For each individual in population:
Randomly select two individuals
Choose the one with better fitness as a parent
Return parents array
Function crossover(parents, crossoverRate, lb,
ub):
Initialize offspring array as parents
For each pair of parents:
If random number is less than crossoverRate:
Perform simulated binary crossover (SBX) to
generate offspring
Apply boundary constraints to offspring
Return offspring
Function mutation(offspring, mutationRate, lb,
ub):
Initialize mutated array as offspring
For each individual in offspring:
If random number is less than mutationRate:
Randomly select a dimension
Perform random mutation within bounds
Return mutated array
Function replaceWorst(population, fitness,
offspring, offspring_fitness):
Combine population and offspring
Combine fitness and offspring fitness
Sort combined fitness
Select the top individuals to form new
population
Return new population and new fitness

IJMSRT25MAY105 www.ijmsrt.com 594

DOI: https://doi.org/10.5281/zenodo.15571838
Volume,3,Issue-5,May2025 International Journal of Modern Science and Research Technology
ISSN NO-2584-2706

2.6 Search Space

The search space illustrated in the benchmark
functions represents diverse optimization
landscapes, including unimodal, multimodal,
convex, and non-convex surfaces. These
functions define the complexity of the
optimization problem, influencing an
algorithm’s ability to explore the space, escape
local optima, and converge toward the global
optimum.

Fig 2. 23 Benchmark Functions

3. Result & Discussion

The performance of the proposed MGO-GA
hybrid approach was evaluated on 23
benchmark functions and compared against
the standard Moss Growth Optimization
(MGO) algorithm. The results indicate that
MGO was outperformed by MGO-GA in 14
out of 23 functions (~61% improvement),
demonstrating that solutions were effectively
refined and optimization performance was
enhanced.

Table 3: Result for 23 Benchmark Function

Function Best Score of Best Score

Name MGO of MGO-
GA
F1 8.2824 6.9672
F2 0.40466 0.70836
F3 8946.4473 8281.6172
F4 27.3275 25.9857
F5 2757.8191 1793.5856
F6 5.7435 4.7054
F7 0.049169 0.029917
F8 -11132.832 -
11793.2468
F9 36.9431 25.2683
F10 1.2067 25.2683
F11 1.0763 1.0718
F12 1.5441 0.61047

IJMSRT25MAY105 www.ijmsrt.com 595

DOI: https://doi.org/10.5281/zenodo.15571838
Volume,3,Issue-5,May2025 International Journal of Modern Science and Research Technology
ISSN NO-2584-2706

F13 5.8758 4.182 [4] D. E. Goldberg, Genetic Algorithms in

F14 0.998 0.998 Search, Optimization & Machine Learning,
Addison-Wesley, 1989.
F15 0.00079911 0.00077964
[5] M. Mafarja et al., “Evolutionary Population
F16 -1.0316 -1.0316 Dynamics and Grasshopper Optimization for
F17 0.39789 0.39955 Feature Selection,” Knowledge-Based Syst., vol.
F18 3 3.0162
145, pp. 25–45, 2018, doi:
10.1016/j.knosys.2017.12.037.
F19 -3.8628 -3.8626 [6] G. Venter and J. Sobieszczanski-Sobieski,
F20 -3.3195 -3.309 “A Genetic Algorithm for Constrained
F21 -10.1406 -10.1455 Optimization Problems,” AIAA J., vol. 40, no. 7,
pp. 1379-1385, 2002.
F22 -10.3868 -10.4029
[7] Y. Liang, A. Törn, and M. Viitanen,
F23 -10.5314 -10.5355 “Benchmark Functions for Global
Optimization,” Springer Optim. Appl., vol. 30,
4. Conclusion pp. 123-150, 2009
The performance of the proposed MGO-GA [8] A. A. Heidari, R. Ali Abbaspour, and A.
hybrid approach was evaluated across 23 Rezaee Jordehi, “An efficient chaotic water
benchmark functions, and comparative cycle algorithm for optimization tasks,” Neural
analysis revealed improvements in 14 of Comput. Appl., vol. 28, no. 1, pp. 57–85, 2017,
them, specifically in functions F1, F3, F4, F5, doi: 10.1007/s00521-015-2037-2.
F6, F7, F8, F9, F11, F12, F13, F15, F21, and [9] R. V. Rao and G. G. Waghmare, “A new
F22. This conclusion was drawn based on optimization algorithm for solving complex
metrics such as solution accuracy, constrained design optimization problems,” vol.
convergence speed, and consistency across 0273, no. April, 2016, doi:
multiple runs. However, certain functions 10.1080/0305215X.2016.1164855.
namely F2, F10, and F19 exhibited minimal [10] R. Al-Hajj and A. Assi, “Estimating
improvement, with F10 showing a noticeable solar irradiance using genetic programming
performance drop. These results indicate that technique and meteorological records,” AIMS
integrating GA operators into MGO enhances Energy, 2017, doi: 10.3934/energy.2017.5.798.
the algorithm's effectiveness in many [11] W. Y. Lin, “A novel 3D fruit fly
scenarios. optimization algorithm and its applications in
economics,” Neural Comput. Appl., 2016, doi:
5. Refrences 10.1007/s00521-015-1942-8.
[12] Y. Cheng, S. Zhao, B. Cheng, S. Hou, Y.
[1] B. Zheng, Y. Chen, C. Wang, A. A. Shi, and J. Chen, “Modeling and optimization
Heidari, L. Liu, and H. Chen, “Moss Growth for collaborative business process towards IoT
Optimization: Concepts and Performance,” applications,” Mob. Inf. Syst., 2018, doi:
Journal of Computational Design and 10.1155/2018/9174568.
Engineering, 2024, doi: [13] X. Wang, T. M. Choi, H. Liu, and X. Yue,
10.1093/jcde/qwae080. “A novel hybrid ant colony optimization
[2] A. A. Heidari, R. Ali Abbaspour, and A. algorithm for emergency transportation problems
Rezaee Jordehi, “An efficient chaotic water during post-disaster scenarios,” IEEE Trans. Syst.
cycle algorithm for optimization tasks,” Man, Cybern. Syst., 2018, doi:
Neural Comput. Appl., vol. 28, no. 1, pp. 57– 10.1109/TSMC.2016.2606440.
85, 2017, doi: 10.1007/s00521-015-2037-2. [14]I. E. Grossmann, Global Optimization in
[3] J. H. Holland, Adaptation in Natural and Engineering Design (Nonconvex Optimization
Artificial Systems, University of Michigan and Its Applications), vol. 9. 1996.
Press, 1975.

IJMSRT25MAY105 www.ijmsrt.com 596

DOI: https://doi.org/10.5281/zenodo.15571838
Volume,3,Issue-5,May2025 International Journal of Modern Science and Research Technology
ISSN NO-2584-2706

[15] E.-S. M. El-Kenawy, M. M. Eid,

M. Saber, and A. Ibrahim, “MbGWO-
SFS: Modified Binary Grey Wolf
Optimizer Based on Stochastic Fractal
Search for Feature Selection,” IEEE
Access, 2020, doi:
10.1109/access.2020.3001151.

IJMSRT25MAY105 www.ijmsrt.com 597

DOI: https://doi.org/10.5281/zenodo.15571838