Browsing by Author "Eker, Erdal"

Now showing 1 - 9 of 9

Comparison of Swarm-Based Metaheuristic and Gradient Descent-Based Algorithms in Artif Icial Neural Network Training
(Ediciones Univ Salamanca, 2023) Eker, Erdal; Kayri, Murat; Ekinci, Serdar; Izci, Davut
This paper aims to compare the gradient descent-based algorithms under classical training model and swarm-based metaheuristic algorithms in feed forward backpropagation artificial neural network training. Batch weight and bias rule, Bayesian regularization, cyclical weight and bias rule and Levenberg-Marquardt algorithms are used as the classical gradient descentbased algorithms. In terms of the swarm-based metaheuristic algorithms, hunger games search, gray wolf optimizer, Archimedes optimization, and the Aquila optimizer are adopted. The Iris data set is used in this paper for the training. Mean square error, mean absolute error and determination coefficient are used as statistical measurement techniques to determine the effect of the network architecture and the adopted training algorithm. The metaheuristic algorithms are shown to have superior capability over the gradient descent-based algorithms in terms of artificial neural network training. In addition to their success in error rates, the classification capabilities of the metaheuristic algorithms are also observed to be in the range of 94%-97%. The hunger games search algorithm is also observed for its specific advantages amongst the metaheuristic algorithms as it maintains good performance in terms of classification ability and other statistical measurements.
Improved Manta Ray Foraging Optimization Using Opposition-Based Learning for Optimization Problems
(Ieee, 2020) Izci, Davut; Ekinci, Serdar; Eker, Erdal; Kayri, Murat
Manta ray foraging optimization (MRFO) algorithm is a bio-inspired meta-heuristic algorithm. It has been proposed as an alternative optimization approach for real-world engineering problems. However, MRFO is not good at fine-tuning of solutions around optima and suffers from slow convergence speed because of its stochastic nature. It needs to be improved due to latter issues. Therefore, in this study, opposition-based learning (OBL) technique was used together with MRFO in order to obtain an effective structure for optimization problems. The proposed structure has been named as opposition-based Manta ray foraging optimization (OBL-MRFO). In the proposed algorithm, the advantage of OBL in terms of considering the opposite solutions was used to have an algorithm with better performance. The proposed algorithm has been tested on four different benchmark functions such as Sphere, Rosenbrock, Schwefel and Ackley. Statistical analyses were performed through comparing the performance of OBL-MRFO with the other algorithms such as salp swarm algorithm, atom search optimization and original MRFO. The results showed that the proposed algorithm is more effective and has better performance than other algorithms.
A New Fusion of Aso With Sa Algorithm and Its Applications To Mlp Training and Dc Motor Speed Control
(Springer Heidelberg, 2021) Eker, Erdal; Kayri, Murat; Ekinci, Serdar; Izci, Davut
An improved version of atom search optimization (ASO) algorithm is proposed in this paper. The search capability of ASO was improved by using simulated annealing (SA) algorithm as an embedded part of it. The proposed hybrid algorithm was named as hASO-SA and used for optimizing nonlinear and linearized problems such as training multilayer perceptron (MLP) and proportional-integral-derivative controller design for DC motor speed regulation as well as testing benchmark functions of unimodal, multimodal, hybrid and composition types. The obtained results on classical and CEC2014 benchmark functions were compared with other metaheuristic algorithms, including two other SA-based hybrid versions, which showed the greater capability of the proposed approach. In addition, nonparametric statistical test was performed for further verification of the superior performance of hASO-SA. In terms of MLP training, several datasets were used and the obtained results were compared with respective competitive algorithms. The results clearly indicated the performance of the proposed algorithm to be better. For the case of controller design, the performance evaluation was performed by comparing it with the recent studies adopting the same controller parameters and limits as well as objective function. The transient, frequency and robustness analysis demonstrated the superior ability of the proposed approach. In brief, the comparative analyses indicated the proposed algorithm to be successful for optimization problems with different nature.
A Novel Enhanced Metaheuristic Algorithm for Automobile Cruise Control System
(Aves, 2021) Izci, Davut; Ekinci, Serdar; Kayri, Murat; Eker, Erdal
The development of a novel enhanced metaheuristic algorithm is considered in this paper. Such a structure was achieved through enhancement of the arithmetic optimization algorithm by employing the opposition-based learning mechanism together with the Nelder-Mead simplex search method. The developed algorithm (ObAOANM) adopts the opposition-based learning scheme to enhance the algorithm in terms explorative behavior, and the Nelder-Mead method in terms of exploitative behavior. The developed ObAOANM was firstly tested against well-known unimodal and multimodal benchmark functions through comparisons with the original arithmetic optimization algorithm, as it was previously shown to be superior to other efficient algorithms. The benchmark functions and related statistical results demonstrated greater capability of the ObAOANM algorithm. Then, the ObAOANM algorithm was utilized to achieve an optimum design of a proportional-integral-derivative controller adopted in an automobile cruise control system. The performance of the ObAOANM algorithm was compared with the arithmetic optimization algorithm algorithm through statistical, transient response, frequency response, and disturbance rejection analyses, which have shown better capability of the enhanced ObAOANM algorithm. Furthermore, the capability of the ObAOANM based proportional-integral-derivative-controlled automobile cruise control system was compared with other available approaches in the literature by performing time domain analysis, which also confirmed the superior capability of the proposed approach for such a task.
A Novel Improved Arithmetic Optimization Algorithm for Optimal Design of Pid Controlled and Bode's Ideal Transfer Function Based Automobile Cruise Control System
(Springer Heidelberg, 2022) Izci, Davut; Ekinci, Serdar; Kayri, Murat; Eker, Erdal
This paper considers the development of a novel hybrid metaheuristic algorithm which is proposed to achieve an optimum design for automobile cruise control (ACC) system by using a proportional-integral-derivative (PID) controller based on Bode's ideal transfer function. The developed algorithm (AOA-NM) adopts one of the recently published metaheuristic algorithms named the arithmetic optimization algorithm (AOA) to perform explorative task whereas another well-known local search method known as Nelder-Mead (NM) simplex search to perform exploitative task. The developed hybrid algorithm was initially tested on well-known benchmark functions by comparing the results with only its original version since AOA has already been shown to be better than other state-of-the-art algorithms. The statistical results obtained from benchmark functions have demonstrated better capability of AOA-NM. Furthermore, a PID controller based on Bode's ideal transfer function was adopted to regulate an ACC system optimally. Statistical, convergence rate, time domain and frequency domain analyses were performed by comparing the performance of AOA-NM with AOA. The respective analyses have shown better capability of the proposed hybrid algorithm. Moreover, the capability of the proposed AOA-NM based PID control scheme was compared with other available approaches in the literature by using time domain analysis. The latter case has also confirmed enhanced capability of the proposed approach for regulating an ACC system which further verified the ability of the proposed AOA-NM algorithm. Lastly, other recently reported and effective metaheuristic algorithms were also used to assess the performance of the proposed approach. The obtained comparative results further confirmed the AOA-NM to be a greater tool to achieve more successful results for ACC system.
Performance Evaluation of Pdo Algorithm Through Benchmark Functions and Mlp Training
(Aves, 2023) Eker, Erdal; Kayri, Murat; Ekinci, Serdar; Kacmaz, Mehmet Ali
Metaheuristic algorithms have become very common in the last two decades. The flexibility and ability to overcome obstacles in solving global problems have increased the use of metaheuristic algorithms. In the training of multilayer perceptron (MLP), metaheuristic algorithms have been preferred for many years due to their good classification capabilities and low error values. Therefore, this study evaluates the performance of the Prairie dog optimization (PDO) algorithm for MLP training. In this context, there are two main focuses in this study. The first one is to test the performance of the PDO algorithm through test functions and to compare it with different metaheuristic algorithms for demonstration of its superiority, and the second is to train MLP using the IRIS dataset with the PDO algorithm. As the PDO is one of the most recent metaheuristic algorithms, the lack of any study on this subject is the motivation for the article. PDO algorithm can be used in real-world problems as a powerful optimizer, as it reaches the minimum point in functions, and can also be used as a classification algorithm because it has successfully performed in MLP training.
Scanm: a Novel Hybrid Metaheuristic Algorithm and Its Comparative Performance Assessment
(Aves, 2022) Kayri, Murat; Ipek, Cengiz; Izci, Davut; Eker, Erdal
This paper proposes a novel sine-cosine and Nelder-Mead (SCANM) algorithm which hybridizes the sine-cosine algorithm (SCA) and Nelder-Mead (NM) local search method. The original version of SCA is prone to early convergence at the local minimum. The purpose of the SCANM algorithm is to overcome this issue. Thus, it aims to overcome this issue with the employment of the NM method. The SCANM algorithm was firstly compared with the SCA algorithm through 23 well-known test functions. The statistical assessment confirmed the better performance of the proposed algorithm. The comparative convergence profiles further demonstrated the significant performance improvement of the proposed SCANM algorithm. Besides, a non-parametric test was performed, and the results that showed the ability of the proposed approach were not by coincidence. A popular and well-performed metaheuristic algorithm known as grey wolf optimization was also used along with the recent and promising two other algorithms (Archimedes optimization and Harris hawks optimization) to comparatively demonstrate the performance of the SCANM algorithm against well-known classical benchmark functions and CEC 2017 test suite. The comparative assessment showed that the SCANM algorithm has promising performance for optimization problems. The non-parametric test further verified the better capability of the proposed SCANM algorithm for optimization problems.
Training Multi-Layer Perceptron Using Harris Hawks Optimization
(Ieee, 2020) Eker, Erdal; Kayri, Murat; Ekinci, Serdar; Izci, Davut
In this paper, Harris hawks optimization (HHO) algorithm has been proposed as an up-to-date meta-heuristic algorithm for training multi-layer perceptron (MLP). The performance of the HHO-based MLP trainer was tested by employing five standard data sets (XOR, Balloon, Iris, Breast Cancer and Heart). The results were compared with those obtained with the sine cosine algorithm (SCA). Comparative statistical results showed that using HHO algorithm as a trainer is more effective and has a higher rate of classification ability.
Training of Artificial Neural Networks Via Modern Meta-Huristic Algorithms
(2020) Eker, Erdal; Kayri, Murat; Ekinci, Serdar
Bu tezin amacı, doğrusal olmayan problemlerin optimizasyonunda kullanılan ve iyi bir sınıflandırma aracı olan Yapay Sinir Ağlarının (YSA) eğitimini sağlamaktır. Tezin amacına uygun olarak YSA mimarilerinden Çok Katmanlı Algılayıcı (Multilayer Perceptron -MLP) kullanılmıştır. MLP eğitiminde olasılıksal algoritma çeşidi olan modern sezgisel-üstü algoritmalar eğitmen olarak seçilmiştir. Algoritmaların genel performansını ölçmek ve performans üstünlüğünü karşılaştırmak için Ackley, Easom, Egg Crate, Griewank, Quartic, Rosenbrock, Schwefel, Sphere ve Step kriter fonksiyonları kullanılmıştır. MLP eğitimi için UCI veri bankasından XOR, İris, Meme Kanseri, Kalp Hastalıkları ve Balloon veri setleri alınmıştır. Tezde dört uygulama yapılmıştır. Uygulamalardan ilki Harris Şahinleri Optimizasyonu sezgisel-üstü algoritmasının (HHO) diğer algoritmalarla karşılaştırılması yapılarak üstünlükleri gösterilmiştir. İkinci uygulamada modern sezgisel-üstü algoritmaların global kısıtlı optimizasyon problemlerinin çözümü gösterilerek Atom Arama Optimizasyonu'nun (ASO) performans üstünlüğü istatistiki olarak ispat edilmiştir. Üçüncü uygulamada MLP eğitimi yapılarak Sinüs Kosinüs Algoritması (SCA) ve HHO algoritmaları karşılaştırılmıştır. Son uygulamada ise önerilen hibrid ASO-SA algoritması (hASO-SA) ile problem çözümündeki sezgisel-üstü algoritmaların dezavantajları ortadan kaldırılmıştır. Performans avantajı istatistiki olarak gösterilerek MLP eğitiminde üstün sınıflandırma becerisi, diğer sezgisel-üstü algoritmalarla karşılaştırılarak açıklanmıştır.