Improved Antlion Optimization Algorithm for Quadratic Assignment Problem

Abstract

The Antlion Optimization (ALO) algorithm is a meta-heuristic optimization algorithm based on the hunting of ants by antlions. The basic inadequacy of this algorithm is that it has long run time because of the random walk model used for the ant's movement. We improved some mechanisms in ALO algorithm, such as ants' random walking, reproduction, sliding ants towards antlion, elitism, and selection procedure. This proposed algorithm is called Improved Antlion Optimization (IALO) algorithm. To show the performance of the proposed IALO algorithm, we used different measurement metrics, such as mean best, standard deviation, optimality, accuracy, CPU time, and number of function evaluations (NFE). The proposed IALO algorithm was tested for different benchmark test functions taken from the literature. There are no studies regarding time analysis of ALO algorithm found in the literature. This study firstly aims to demonstrate the success of the proposed IALO algorithm especially in runtime analysis. Secondly, the IALO algorithm was also applied to the Quadratic Assignment Problem (QAP) known as a difficult combinatorial optimization problem. In QAP tests, the performance of the IALO algorithm was compared with the performances of the classic ALO algorithm and 14 well-known and recent meta-heuristic algorithms. The results of the benchmark test functions show that IALO algorithm is able to provide very competitive results in terms of mean best/standard deviation, optimality, accuracy, CPU time and NFE metrics. The CPU time results prove that IALO algorithm is faster than the classic ALO algorithm. As a result of the QAP tests, the proposed IALO algorithm has the best performance according to the mean cost, worst cost and standard deviation. The source codes of QAP with the proposed IALO algorithm are publicly available at https://github.com/uguryuzgec/QAP-with-IALO.

A Hybrid Method for Optimal Arrangement of Turbine Blades

“…Therefore, researchers are seeking meta-heuristic algorithms to solve them in a reasonable time. For examples of these methods, one may address the ant colony [14,15], the hybrid particle swarm optimization with simulated annealing [16], the antlion optimization algorithm [17], and the whale optimization algorithm [18].…”

Re: “Dealing With The Problem Of Planning Construction Site Layout Based On Quadratic Assignment Problem Method, Using Ant-Lion hybrid algorithm.

“…After hunting, the Antlions update their position with the position of the ants according to the fitness value using the following equation: 2.2. Improved ALO (IALO) [26] Reducing the size of the random walk in the population ant at the start of (ALO) algorithm, the exploration and exploitation of (ALO) below 20% in the maximum iteration, the 20% maximum iteration change in ant's random walk model is shown in the following code snippet: (𝑡) = [0, ⋯ , 𝑐𝑢𝑚𝑠𝑢𝑚(2𝑟(𝑡𝑛) -1)], 𝑛 = 1,2, ⋯ , 𝑀𝑎𝑥_𝑖𝑡𝑒𝑟/5 for every ant (16) Improved selection of Antlion in roulette wheel by increasing fitness value and choosing negative fitness value, at the end of algorithm, update the elitist Antlion, combine search and sort population, compare fitness value pairs of ant and Antlion, if the fitness value of ant is better than the fitness value of Antlion, the position of ant will be updated to Antlion as shown by the following code: Select Antlion by roulette wheel method for building trap [27] Tournament selection is a simple selection method, n individuals are selected from a large population, and left for tournament. The individual with the highest fitness will be selected, each selection is usually two individuals (binary tournament) and sorted into a new group.…”

An Improved Ant-Lion Optimization for Deep and Handcrafted Feature Selection in MR Brain Tumor Classification

“…he weights of the sub-goals are specified by λ ∈ [0, 1]. The optimization problem has as many dimensions (variables) as there are features [46], and each variable will always range between [0, 1]. To stop termination, two criteria were used: one criteria verifies the exceeding the maximum number of iterations T max ,and other criteria checks for Fitness value to reach ( FVR = 10…”