Ahmet Babalik

Accelerated ABC (A-ABC) Algorithm for Continuous Optimization Problems

In this paper, accelerated artificial bee colony (A-ABC) method is presented. In A-ABC, two modifications are used on the Artificial Bee Colony (ABC) algorithm to progress its local search ability and convergence speed. Modifications are called as modification rate (MR) and step size (SS). In this study, we aim to investigate effects of using MR value and SS modification. Performances of the A-ABC and standard ABC are compared on well-known 7 different benchmark functions. Results show that A-ABC has generally better performance and faster convergence than standard version of the ABC algorithm. Index Terms—ABC algorithm, meta-heuristic algorithms, swarm intelligence, optimization technique.

Classifiers fusion in recognition of wheat varieties

Five wheat varieties (Bezostaja, Cesit1252, Daǧdas, Gerek, Kiziltan traded in Konya Exchange of Commerce, Turkey), characterized by nine geometric and three colour descriptive features have been classified by multiple classier system where pair-wise SLP or SV classifiers served as base experts. In addition to standard voting and Hastie and Tibshirani fusion rules, two new ones were suggested that allowed reducing the generalization error up to 5%. In classifying of kernel lots, we may obtain faultless grain recognition.

An improvement in fruit fly optimization algorithm by using sign parameters

The fruit fly optimization algorithm (FOA) has been developed by inspiring osphresis and vision behaviors of the fruit flies to solve continuous optimization problems. As many researchers know that FOA has some shortcomings, this study presents an improved version of FOA to remove with these shortcomings in order to improve its optimization performance. According to the basic version of FOA, the candidate solutions could not take values those are negative as well as stated in many studies in the literature. In this study, two sign parameters are added into the original FOA to consider not only the positive side of the search space, but also the whole. To experimentally validate the proposed approach, namely signed FOA, SFOA for short, 21 well-known benchmark problems are considered. In order to demonstrate the effectiveness and success of the proposed method, the results of the proposed approach are compared with the results of the original FOA, results of the two different state-of-art versions of particle swarm optimization algorithm, results of the cuckoo search optimization algorithm and results of the firefly optimization algorithm. By analyzing experimental results, it can be said that the proposed approach achieves more successful results on many benchmark problems than the compared methods, and SFOA is presented as more equal and fairer in terms of screening the solution space.

A modification of tree-seed algorithm using Deb’s rules for constrained optimization

Abstract This study focuses on the modification of Tree-Seed Algorithm (TSA) to solve constrained optimization problem. TSA, which is one of the population-based iterative search algorithms, has been developed by inspiration of the relations between trees and seeds grown on a land, and the basic version of TSA has been first used to solve unconstrained optimization problems. In this study, the basic algorithmic process of TSA is modified by using Deb’s rules to solve constrained optimization problems. Deb’s rules are based on the objective function and violation of constraints and it is used to select the trees and seeds that will survive in next iterations. The performance of the algorithm is analyzed under different conditions of control parameters of the proposed algorithm, CTSA for short, and well-known 13 constrained maximization or minimization standard benchmark functions and engineering design optimization problems are employed. The results obtained by the CTSA are compared with the results of particle swarm optimization (PSO), artificial bee colony algorithm (ABC), genetic algorithm (GA) and differential evolution (DE) algorithm on the standard benchmark problems. The results of state-of-art methods are also compared with the proposed algorithm on engineering design optimization problems. The experimental analysis and results show that the proposed method produces promising and comparable results for the constrained optimization benchmark set in terms of solution quality and robustness.

A Pre-Processing Approach Based on Artificial Bee Colony for Classification by Support Vector Machine

In this study, handling with the success of pre-processing on classification tasks, artificial bee colony (ABC) algorithm is used as a pre-processor in order to improve accuracy of the support vector machine (SVM) classifier. Proposed approach is examined on three different online available dataset by using k-fold cross validation method. The results obtained are compared with the results of the classification of the datasets with pure SVM classifier. The increase of the classification accuracy is observed. By altering parameters of the suggested approach, it is thought the approach would be more successful on the different datasets.

An artificial algae algorithm for solving binary optimization problems

This paper focuses on modification of basic artificial algae algorithm (AAA) for solving binary optimization problems by using a new solution update rule because the agents in AAA work on continuous solution space. The candidate solution generation process of algorithm in the basic version of AAA is replaced with a mechanism that use a neighbor solution randomly selected from the population and three decision variables of this solution. The current solution is taken from the population and randomly selected three dimensions of this solution are changed using the neighbor solution. The agents of AAA work on continuous solution space and this modification for AAA is required for solving a binary optimization problem because a binary optimization problems have decision variables which are element of set {0, 1}. The performance of the proposed algorithm, binAAA for short, is investigated on the uncapacitated facility location problems which are pure binary optimization problem and there is no integer or real valued decision variables in this problem. The results obtained by binAAA are compared with the results of state-of-art algorithms such as artificial bee colony, particle swarm optimization, and genetic algorithms. Experimental results and comparisons show that the binAAA is better than the other algorithm almost all cases in terms of solution quality and robustness based on the mean and standard deviations, respectively.

A Multi-Objective Artificial Algae Algorithm

Abstract In this study, the authors focus on modification of the artificial algae algorithm (AAA), for multi-objective optimization. Basically, AAA is a population-based optimization algorithm inspired by the behavior of microalgae cells. In this work, a modified AAA with appropriate strategies is proposed for multi-objective Artificial Algae Algorithm (MOAAA) from the first AAA that was initially presented to solve single-objective continuous optimization problems. To the best of our knowledge, the MOAAA is the first modification of the AAA for solving multi-objective problems. Performance of the proposed algorithm is examined on a benchmark set consisting of 36 different multi-objective optimization problems and compared with four different swarm intelligence or evolutionary algorithms that are well-known in literature. The MOAAA is highly successful in solving multi-objective problems, and it has been demonstrated that the MOAAA is an alternative competitive algorithm in multi-objective optimization according to experimental results and comparisons presented in this research topic.

Implementation of Bat Algorithm on 2D Strip Packing Problem

This paper suggests utilization of a novel metaheuristic method namely bat algorithm (BA) in order to solve 2D rectangular strip packing problem. Although BA is proposed for solving continuous optimization problems, a discrete version of BA is developed by being used neighborhood operators to solve the problem dealt with this study. Firstly, bottom left approach is used as the placement algorithm in the problem, then, discrete BA is used for obtaining the proper sequence of the rectangular object list. The performance of the proposed approach is investigated on 9 different problems on well-known 2D rectangular problem literature. Experimental results show that discrete BA is effective and alternatively usable in solving 2D rectangular strip packing problems.

Effects of Feature Selection Using Binary Particle Swarm Optimization on Wheat Variety Classification

In this article, classification of wheat varieties is aimed with the help of multiclass support vector machines (M-SVM) and binary particle swarm optimization (BPSO) algorithm. For each wheat kernel, 9 geometric and 3 color features are obtained from the digital images which are belong to 5 wheat type. Wheat types are classified using M-SVM. In order to increase the reliability of the classification process, 2 fold cross validation approach is implemented and this process repeated 250 times. Average classification accuracy is obtained as 91.5%. With the aim of increasing the classification accuracy and decreasing the process time, descriptive features are decreased by BPSO algorithm and reduced from 12 to 7. Average of classification accuracy is obtained as 92.02% using 7 features obtained from reduction with BPSO.

Recognition of Sun-pest Infected Wheat Kernels Using Artificial Neural Networks

In this study it is aimed to recognize sun-pest infected kernels in a sample sub-group of wheat kernels taken from a bulk of Bezostaja wheat. Recognition of the damaged kernels is realized by evaluating light transmittance data of the kernels through use of artificial neural networks (ANN). Wheat kernels in the sub-group are left to fall in an oblique groove with semi-circular cross-section. While the kernels cross a LED light source, light transmitted through the kernel fall on a sensor just across the light source. Analog signals induced by the sensor are recorded and histograms of these signals are evaluated by using ANN in order to recognize sun-pest infected wheat kernels in the sub-group. Two different ANN models: multi layer perceptron (MLP) and self organizing map (SOM) models were used in the recognition process.