Experimental Study of Swarm Migration Algorithms on Stochastic and Global Optimisation Problem

Experimental Study of Swarm Migration Algorithms on Stochastic and Global Optimisation Problem

Stephen Opoku Oppong, Benjamin Ghansah, Evans Baidoo, Wilson Osafo Apeanti, Daniel Danso Essel
Copyright: © 2022 |Pages: 26
DOI: 10.4018/IJDAI.296389
(Individual Articles)
No Current Special Offers


Complex computational problems are occurrences in our daily lives that needs to be analysed effectively in order to make meaningful and informed decision. This study performs empirical analysis into the performance of six optimisation algorithms based on swarm intelligence on nine well known stochastic and global optimisation problems, with the aim of identifying a technique that returns an optimum output on some selected benchmark techniques. Extensive experiments show that, Multi-Swarm and Pigeon inspired optimisation algorithm outperformed Particle Swarm, Firefly and Evolutionary optimizations in both convergence speed and global solution. The algorithms adopted in this paper gives an indication of which algorithmic solution presents optimal results for a problem in terms of quality of performance, precision and efficiency.
Article Preview


According to Pham & Karaboga (2000), solutions to most optimization problems cannot be obtained by traditional mathematical programming approaches. They opined that present heuristic algorithms have been created for solving combinatorial and numeric optimization problems to obtain universal optimum solutions. Optimization problems are pervasive in our daily lives and usually take different forms. Generally speaking, optimization problems can be found in broad applications such as finance and logistics, manufacturing, and engineering. For example, logistic regression classification prepares classifiers in a unique situation as it tries to discover a set of values for the weights related to the input variables so that for a set of training data, the calculated output values directly match the identified output variable values. Over the past decade, there has been tremendous development in countless optimization algorithms in varied communities that cross intersect various fields that include artificial intelligence, computational intelligence, and soft computing (Gholizadeh & Barati, 2012).

Inspired by nature, several successful algorithms are suggested and categorized into diverse groups depending on the measure being considered, thus, stochastic, iterative based, population-based or deterministic, with examples as, Artificial Fish Swarm Algorithm (Li, 2002; Tan & Zhu, 2010), Flower Pollination Algorithm (Yang, 2012), Differential Evolution (Das & Suganthan, 2011), Firefly Algorithm (Yang, 2009), imperialist competitive algorithm (Atashpaz-Gargari & Lucas, 2007), brainstorm optimization (Shi, 2011) among others. Among the listed criteria, a population-based algorithm works with a set of solutions with a specific aim of trying to improve upon them, whereas an iterative algorithm uses multiple iterations to search after a solution. On the other hand, when an algorithm makes use of a probabilistic rule for recovering a solution then it is said to be stochastic.

These nature-inspired algorithms with their simple structure, fewer adjustment to their parameters, and their ability to obtain optimum results, have over the years drawn the attention of most researchers to apply them to diverse fields of research including mechanic control, machine engineering, software checking, traditional combinatorial problems, and cluster investigations. Some practical applications of these nature-inspired algorithms include intrusion detection (Srivastava et al., 2021), the medical field (Gadekallu et al., 2020), machine learning, and deep learning (Abidi et al., 2021), IoT networks (Alazab et al., 2021) among others. Different proposals of nature-inspired algorithms for different purposes have created great confusion as to which method fits what situation, and it is strongly required to understand what the common characteristics are of these algorithms and what the differences are. To study and compare the characteristics of these algorithms can provide not only a broader perspective to the improvement of the current nature-inspired algorithms but also a solid and feasible cornerstone for building up the new problem-oriented nature-inspired algorithms (Wang et al., 2021).

Nature-inspired algorithms can be classified as those based on biology and those inspired by natural phenomena. The algorithms based on biology can be further divided into those based on evolution and those based on swarm behavior. The evolutionary algorithms include genetic algorithms, differential evolution, cultural evolution, evolutionary strategies, and genetic programming. The swarm category includes particle swarm optimization, ant colony optimization, artificial bees, termites, bats, birds, cats, bacterial foraging, cuckoo search, and firefly algorithm (Valdez et al., 2021). Bio-inspired algorithms form a majority of all nature-inspired algorithms. From the set theory point of view, the swarm intelligence-based algorithms are a subset of bio-inspired algorithms, while the bio-inspired algorithms are a subset of nature-inspired algorithms.

Complete Article List

Search this Journal:
Volume 15: 1 Issue (2024): Forthcoming, Available for Pre-Order
Volume 14: 2 Issues (2022)
Volume 13: 2 Issues (2021)
Volume 12: 2 Issues (2020)
Volume 11: 2 Issues (2019)
Volume 10: 2 Issues (2018)
View Complete Journal Contents Listing