Bilgisayar ve Bilişim Fakültesi Koleksiyonu

Permanent URI for this collectionhttps://hdl.handle.net/20.500.13091/10834

Browse

Now showing 1 - 19 of 19

Citation - WoS: 7
Citation - Scopus: 12
Analyzing the Effect of Data Preprocessing Techniques Using Machine Learning Algorithms on the Diagnosis of Covid-19
(Wiley, 2022) Erol, Gizemnur; Uzbaş, Betül; Yücelbaş, Cüneyt; Yücelbaş, Sule
Real-time polymerase chain reaction (RT-PCR) known as the swab test is a diagnostic test that can diagnose COVID-19 disease through respiratory samples in the laboratory. Due to the rapid spread of the coronavirus around the world, the RT-PCR test has become insufficient to get fast results. For this reason, the need for diagnostic methods to fill this gap has arisen and machine learning studies have started in this area. On the other hand, studying medical data is a challenging area because the data it contains is inconsistent, incomplete, difficult to scale, and very large. Additionally, some poor clinical decisions, irrelevant parameters, and limited medical data adversely affect the accuracy of studies performed. Therefore, considering the availability of datasets containing COVID-19 blood parameters, which are less in number than other medical datasets today, it is aimed to improve these existing datasets. In this direction, to obtain more consistent results in COVID-19 machine learning studies, the effect of data preprocessing techniques on the classification of COVID-19 data was investigated in this study. In this study primarily, encoding categorical feature and feature scaling processes were applied to the dataset with 15 features that contain blood data of 279 patients, including gender and age information. Then, the missingness of the dataset was eliminated by using both K-nearest neighbor algorithm (KNN) and chain equations multiple value assignment (MICE) methods. Data balancing has been done with synthetic minority oversampling technique (SMOTE), which is a data balancing method. The effect of data preprocessing techniques on ensemble learning algorithms bagging, AdaBoost, random forest and on popular classifier algorithms KNN classifier, support vector machine, logistic regression, artificial neural network, and decision tree classifiers have been analyzed. The highest accuracies obtained with the bagging classifier were 83.42% and 83.74% with KNN and MICE imputations by applying SMOTE, respectively. On the other hand, the highest accuracy ratio reached with the same classifier without SMOTE was 83.91% for the KNN imputation. In conclusion, certain data preprocessing techniques are examined comparatively and the effect of these data preprocessing techniques on success is presented and the importance of the right combination of data preprocessing to achieve success has been demonstrated by experimental studies.
Binary African Vultures Optimization Algorithm for Z-Shaped Transfer Functions
(2023) Baş, Emine
Metaheuristic algorithms are of great importance in solving binary optimization problems. African Vulture Optimization algorithm (AVO) is a swarm intelligence-based heuristic algorithm created by imitating the life forms of African vultures. In this study, the AVO, which has been proposed in recent years, is restructured to solve binary optimization problems. Thus, Binary AVO (BAVO) has been proposed. Four different z-shaped transfer functions are chosen to convert the continuous search space to binary search space. Variations for BAVO are defined according to the transfer function used (BAVO1, BAVO2, BAVO3, and BAVO4). The success of these variations was tested in thirteen classic test functions containing unimodal and multimodal functions. Three different dimensions were determined in the study (5, 10, and 20). Each test function was run ten times independently and the average, standard deviation, best, and worst values were obtained. According to the results obtained, the most successful of these variations has been identified. According to the results, the BAVO4 variant at higher dimensions achieved better results. The success of BAVO with z-shaped transfer functions was demonstrated for the first time in this study.
Binary Fox Optimization Algorithm Based U-Shaped Transfer Functions for Knapsack Problems
(2023) Baş, Emine
This paper examines a new optimization algorithm called Fox optimizer (FOX), which mimics the foraging behavior of foxes while hunting in nature. When the literature is examined, it is seen that there is no version of FOX that solves binary optimization problems. In this study, continuous search space is converted to binary search space by U-shaped transfer functions and BinFOX is proposed. There are four U-shaped transfer functions in the literature. Based on these transfer functions, four BinFOX variants are derived (BinFOX1, BinFOX2, BinFOX3, and BinFOX4). With BinFOX variants, 25 well-known 0-1knapsack problems in the literature have been solved and their success has been demonstrated. The best, worst, mean, standard deviation, time, and gap values of each variant were calculated. According to the results, the most successful BinFOX variant was determined. The success of BinFOX with U-shaped transfer functions was demonstrated for the first time in this study.
Business Strategy and Market-Based Performance
(2023) Baş, Emine
Market orientation, which is one of the most remarkable orientations; It is the whole of organizational activities aimed at understanding and satisfying the general demands and needs of customers and providing unique customer value. However, in a rapidly changing competitive environment, there is a need for competitive tactics that will strengthen the market orientation and directly contribute to performance, rather than focusing only on market orientation. In this context, the relationships between the components of market orientation, differentiation strategy and firm performance are of great importance.
Büyük Veri ve Hadoop
(2022) Baş, Emine
Günümüzde teknolojinin yaygın bir şekilde kullanılmasıyla artan bir very (büyük veri) oluşmuştur. Büyük veri, geleneksel veri işleme uygulamalarının üstesinden gelemeyeceği kadar büyük veya karmaşık veri setlerini analiz etme ve bu veri setlerinden sistematik olarak bilgi elde etmeyi sağlayacak yöntemler arayan bilişim bilimleri sahasıdır. Bir diğer deyişle Big Data, çoğunluğu yapılandırılmamış olan ve sonu gelmez bir şekilde birikmeye devam eden, geleneksel ilişki bazlı veri tabanı teknikleri yardımıyla çözülemeyecek kadar yapısallıktan uzak, çok çok büyük, çok ham ve üstel bir şekilde büyümekte olan veri setleridir. Büyük very çeşitlilik, hız ve hacim olmak üzere üç ana bileşeni ile karakterize edilen geleneksel veri analizinden devrim niteliğinde bir adım gerektirir. Bu verinin şekli itibariyle klasik yöntemlerle işlenmesi zordur. Çeşitlilik (Variety), büyük verileri gerçekten büyük hale getirir. Verinin hacmi veya boyutu (Volume) artık terabayt ve petabayttan daha büyüktür. Hız (Velocity) sadece büyük veri için değil, tüm süreçler için gereklidir. Zaman sınırlı süreçler için, değerini en üst düzeye çıkarmak için kuruluşa akarken büyük veri kullanılmalıdır. Verilerin büyük ölçeği ve yükselişi, geleneksel depolama ve analiz tekniklerini geride bırakır. Araştırmacılar bu verinin kolay bir şekilde işlenmesi için bir arayış içine girmiştir. Büyük veri, MapReduce gibi mimarileri destekleyen yepyeni bir endüstri yaratmıştır. Hadoop bu büyük verinin sınıflandırılması ve işlenmesi konusunda çıkmış bir yazılımdır. Hadoop JAVA programlama dili ile geliştirilmiş popüler, açık kaynaklı bir Apache projesidir. Üretilme amacı ise büyük verilerin daha hızlı işlenmesidir. Temel olarak yazılımı dağıtık dosya sistemi olarak tanımlayabiliriz. Bu dağıtık dosya sistemi HDFS yani Hadoop Distributed File System olarak adlandırılır. Hadoop bileşenleri şunlardır: HDFS, MapReduce, HBase, Pig, Hive ve ZooKeeper dir. Bu bildiride büyük veri ve hadoop konusunda bir araştırma sunulmuştur.
Citation - WoS: 15
Citation - Scopus: 19
A Comprehensive Analysis of Grid-Based Wind Turbine Layout Using an Efficient Binary Invasive Weed Optimization Algorithm With Levy Flight
(Pergamon-Elsevier Science Ltd, 2022) Koç, İsmail
Wind energy has attracted great attention in recent years due to the increasing demand for alternative energy sources. Gathering the maximum amount of energy from wind energy is directly related to the layout of wind turbines in wind farm. This study focuses on grid-based wind turbine layout problem in an area of 2 km x 2 km. In order to solve this problem, 9 novel grids of 11 x 11, 12 x 12, ... and 19 x 19 are proposed in this paper in addition to 10 x 10 and 20 x 20 turbine grids which have already used in the literature. A new repair operator is recommended, taking into account the distance constraint between two adjacent cells in turbine layout problem, and thus the obtained solutions are made feasible. Furthermore, a levy flight-based IWO (LFIWO) algorithm is developed to optimize the layout of the turbines in wind farm. The basic IWO algorithm and the proposed LFIWO algorithm are compared on 11 different turbine layouts. Experimental studies are carried out for both algorithms under equal conditions with the help of 10 different binary versions. According to the comparisons performed using 11 different grids, LFIWO demonstrates a much superior success than the IWO algorithm. In addition, the proposed 19 x 19 grid reveals the best success among the other grids. When compared to the other studies in the literature, it is seen that LFIWO surpasses the other algorithms in terms of solution quality. As a result, it can be clearly stated that the proposed binary version of LFIWO is a competitive and effective binary algorithm.
Detailed Parameter Analysis for African Vultures Optimization Algorithm
(2023) Baş, Emine
Metaheuristic algorithms are of great importance in solving optimization problems. In this study, the newly proposed African Vulture Optimization algorithm (AVO) has been examined. The AVO algorithm mimics the life-styles of African vultures and was created by imitating the foraging and wandering behavior of African vultures. Six kinds of fixed parameters (P1, P2, P3, L1, L2, w) are used in the algorithm. While the original paper examined the effect of these parameter values on AVO for only six types of values, nine types of effects were examined in this study (L1={0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1}, L2={0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9}, P1={0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9}, P2={0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9}, P3={0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9}, w= {1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5}). The best parameter values were selected for AVO by examining the results. These parameters balance AVO's local and global search capabilities. According to the results, while the values of L1, L2, and w parameters were similar to the values in the original paper (L1=0.6, L2=0.4, and w=2.5), different appropriate values were determined for P1, P2, and P3 values (P1=0.4, P2=0.9, and P3=0.6).
Citation - WoS: 15
Citation - Scopus: 17
Discrete Tree Seed Algorithm for Urban Land Readjustment
(Pergamon-Elsevier Science Ltd, 2022) Koç, İsmail; Atay, Yılmaz; Babaoğlu, İsmail
Land readjustment and redistribution (LR) is an important approach used to realize development plans by converting rural lands to urban land and also providing urban infrastructure. The LR problem, which is a complex challenging real-world problem, is a discrete optimization problem because its structure is similar to TSP (Traveling Salesman Problem) and scheduling problems which are combinatorial optimization problems. Since classical mathematical methods are insufficient for solving NP (Nondeterministic Polynomial) optimization problems due to time limitations, meta-heuristic optimization algorithms are commonly utilized for solving these kinds of problems. In this paper, meta-heuristic algorithms including genetic, particle swarm, differential evolution, artificial bee, and tree seed algorithms are utilized for solving LR problems. The stated meta-heuristic algorithms are used by applying spatial-based crossover and mutation operators depending upon the LR problem on each algorithm. Moreover, a synthetic dataset is used to ensure that the quality of the solution obtained is acceptable to everyone, to prove an optimal solution easily. By utilizing the suggested spatial-based crossover and mutation operators, finding the ideal solution is aimed using the synthetic dataset. In addition, five different modifications on TSA (Tree-Seed Algorithm) are performed and used to solve LR problems. All the modified versions of TSA are carried out only by changing the mechanism of seed reproduction. The novel TSA approaches are respectively named as tcTSA (tournament current), tbTSA (tournament best), pbTSA (personal-best based), t2TSA (double tournament), and elTSA (elitism based). In the experimental studies, the hybrid approach, which includes the crossover and mutation operators, is successfully applied in all of the algorithms under equal conditions for a fair comparison. According to experimental results performed using the dataset, it can be clearly stated that especially t2TSA outperforms all the algorithms in terms of performance and time.
Energy Demand Projection of Turkey Based on Coot Bird Metaheuristic Optimizer
(2021) Koç, İsmail
The energy demand projection forms the basis of realistic energy planning. In this study, from 1979 to 2011, a 33- year data set including gross domestic product (GDP), population, import and export was used for energy demand forecasting in Turkey. Using this data set, two different energy estimation models, linear (COOT_L) and quadratic (COOT_Q), were developed with the Coot bird metaheuristic optimizer. These models were compared with different optimization algorithms in the literature. When the experimental results were examined, the COOT_L model produced a more successful result than the results of other algorithms. Furthermore, COOT_Q was seen to become more successful than PSO and ACO. In addition to these, the COOT_L and COOT_Q models forecasted the future energy demand between 2012 and 2030 in Turkey and their results were compared with those of DE methods. When the results were examined, while DE and COOT_L produced similar results in linear form, certain differences were observed among the results of COOT_Q and DE.
Citation - WoS: 17
Citation - Scopus: 23
A Fast Community Detection Algorithm Based on Coot Bird Metaheuristic Optimizer in Social Networks
(Pergamon-Elsevier Science Ltd, 2022) Koç, İsmail
Community detection (CD) is critical to understanding complex networks. Researchers have made serious efforts to develop efficient CD algorithms in this sense. Since community detection is an NP-hard problem, utilizing metaheuristic algorithms is preferred instead of classical approaches in solving the problem. For this reason, in this study, six different metaheuristic algorithms called Archimedes optimization algorithm (AOA), Atom search optimization (ASO), Coot Bird Natural Life Model (COOT), Harris Hawks Optimization (HHO), Slime Mould Algorithm (SMA) and Arithmetic Optimization Algorithm (AROA) are used in the solution of CD problems and all of which have been proposed for solving continuous problems in recent years. Since the CD problem has a discrete structure, discrete versions of all the algorithms are produced, and then the proposed discrete algorithms are adapted to the problem. In addition, in the phase of evaluating the objective function of the problem, a fast approach based on CommunityID is proposed to minimize the time cost when solving the problem, and this approach is utilized in all the algorithms when calculating the fitness value. In the experimental studies, firstly, the novel discrete algorithms are compared with each other in terms of solution quality and time and according to these results, COOT becomes the most effective and very fast algorithm. Then, the results obtained by COOT are compared with those of important studies in the literature. When compared in terms of solution quality, it is seen that the COOT algorithm is more effective than the other algorithms. In addition, it is quite obvious that all of the proposed algorithms using the CommunityID-based approach are faster than the other algorithms in the literature in terms of time. As a result, it can be said that COOT can be an effective alternative method for dealing with CD problems. In addition, the approach based on CommunityID can also be utilized in larger networks to obtain remarkable solutions in a much shorter time.
Fox Optimization Algorithm for Cec-2017 Benchmarks Problems
(2023) Baş, Emine
In this study, the newly proposed Fox optimization algorithm (FOX) has been studied. The FOX algorithm is an algorithm that imitates the hunting movement of red foxes living in nature in snowy environments. The FOX algorithm was first proposed by Mohammed and Rashid in 2023. They demonstrated the success of FOX in CEC-2019 and engineering design problems. Since FOX is new, its success in different test functions has not been shown in the literature. In this study, the success of FOX is demonstrated in the CEC-2017 test functions, which consist of 29 test functions, which include many different problem types (unimodal, multimodal, hybrid, and composition). Three different sizes (10, 30, and 50) of FOX were run, presenting a variety of results. FOX was run independently 20 times for each CEC-2017 test function. Results are shown according to mean, standard deviation, best, worst, and time comparisons. These results can be used in FOX comparisons in different studies in the literature. The results obtained in this study constitute a source of comparison to other studies using the CEC-2017 test functions.
Identification of Covid-19 From Cough Sounds Using Non-Linear Analysis and Machine Learning
(2021) Solak, Fatma Zehra
Automatic diagnosis of COVID-19 has an active role in reducing the spread of the disease by minimizing interaction with people. Machine learning models using various signals and images form the basis of automatic diagnosis. This study presents the machine learning based models for detecting COVID-19 infection using ‘Virufy’ dataset containing cough sound signals labeled as COVID-19 and Non-COVID-19. Since the number of COVID positive coughs in the set is less than those of COVID negative, firstly, data balancing was performed with the ADASYN oversampling technique in the study. Then, features were extracted by non-linear analysis of cough sounds using Multifractal Detrended Fluctuation Analysis (MDFA), Lempel–Ziv Complexity (LZC) and entropy measures. Later, the most effective features were selected by ReliefF method. Finally, five machine learning algorithms, namely Support Vector Machine with Radial Basis Function (SVM-RBF), Random Forest (RF), Adaboost, Artificial Neural Network (ANN), k Nearest Neighbor (kNN) were used to identify cough sounds as COVID-19 or Non-COVID19. As a result of the study, the cough sounds of COVID-19 patients and Non-COVID19 subjects were identified with 95.8% classification accuracy thanks to the RBF kernel function of SVM and the selected effective features. With this classifier, 93.1% sensitivity, 98.6% specificity, 98.6% precision, 0.92 kappa statistical values and 93.2% area under the ROC curve were obtained.
A New Binary Snake Optimizer for 0-1 Knapsack Problems
(2023) Baş, Emine
The Snake Optimizer (SO) is a newly proposed heuristic algorithm in recent years. It was proposed in the original paper for continuous optimization problems. Its success has been tested on CEC-2017 and engineering design problems. When the literature is reviewed, there is no recommended version of SO for binary optimization problems. In this study, SO has been updated to solve binary optimizations. Transfer functions are generally used when converting continuous search space to binary search space. In this study, S-shaped and Vshaped transfer functions, which are mostly used in binary optimizations, are used. Eight different Binary SO (SOBin) variations were obtained according to the transfer functions used (SOBin_S1, SOBin_S2, SOBin_S3, SOBin_S4, SOBin_V1, SOBin_V2, SOBin_V3, SOBin_V4). These Binary SO variations were tested on twenty-five knapsack problems of different sizes (dimension= {8, 12, 16, 20, 24}). The knapsack problem is often used to test the success of binary optimization problems. The knapsack problem is based on placing the most valuable and least weighty objects in a bag. It is a maximization problem. According to the results obtained, V-shaped transfer functions have obtained more successful results than S-shaped transfer functions. The most successful Binary SO variation was the one using the V1 transfer function (SOBin_V1).
Performance of Snake Optimizer on Cec-C06 2019 Test Functions
(2023) Baş, Emine
In this study, the Snake Optimizer (SO) algorithm, which has been proposed in recent years, has been examined. The SO is a heuristic algorithm proposed for continuous optimization problems. SO is a heuristic algorithm created by imitating the special mating movements of snakes. In its original paper, SO's performance results for CEC-2017 and engineering design problems were examined. In order to prove the success of a heuristic algorithm, it is necessary to demonstrate its success in different benchmark functions. In this study, the success of SO is demonstrated in IEEE CEC-C06 2019 benchmark functions according to three different population sizes (30, 50, and 100) and three different maximum iteration (500, 1000, and 5000). IEEE CEC-C06 2019 test functions are a set of benchmark functions consisting of ten different single-purpose optimization groups. It is often preferred as a problem function set for comparison of continuous optimization problems. Convergence graphs were drawn according to each population size and its success was demonstrated. Thus, a contribution has been made to the literature by showing SO's results in IEEE CEC-C06 2019 benchmark functions. It provided a source for other heuristic algorithms in comparisons.
Privacy Preserving Multi-Proxy Based Encrypted Keyword Search
(2022) Öksüz, Özgür
This paper presents a multi-proxy (2 proxies) based encrypted keyword search scheme that enjoys the following properties: This scheme provides data confidentially that encrypted data does not leak any keywords and documents to the attackers (data server and a proxy). Moreover, the proposed scheme provides trapdoor privacy. In other words, the attackers do not learn any information about searched keywords. Furthermore, even if a proxy is controlled by an attacker, the attacker does not learn any information about the queries (keywords that the user searches over the database) and database results. Different from other studies, this scheme provides lightweight user-side query and data processing. In other words, most of the job (query processing) is done by the proxies on behalf of the user. Finally, the proposed scheme relaxes the trust assumption that eliminates a single point of failure by introducing multi-proxy architecture.
Snake Optimizer for Large-Scale Optimizaton Problems
(2023) Baş, Emine
The Snake Optimizer (SO) is a newly proposed heuristic algorithm in recent years. It was proposed in the original paper for continuous optimization problems. When the literature was reviewed, it was noticed that the success of SO for large-sized problems was not tested. In this study, the success of SO was examined on data sets consisting of six different large-sized (1024, 3072, and 4868) EEG signals, known as the big data optimization problem. The success of SO has been thoroughly investigated on a big data optimization problem in three different iterations (100, 300, and 500) and three different population sizes (30, 50, and 100). The convergence graphs of the problem datasets according to the population size were drawn and examined. SO was run independently twenty times for each dataset. Statistical evaluations such as average, standard deviation, best, worst, and time were made on the results obtained. According to the average results, the population size and the maximum number of iterations have a direct effect on the result, but they also increase the solution time of the problem. SO has been compared with various heuristic algorithms selected from the literature (Jaya, AOA, BA, PSO-Q, and IPSO-Q). According to the results, SO achieved better results in all big data optimization problems. The results showed that the SO heuristic algorithm was able to maintain its success as the size of the problem increased. This comes from SO's ability to explore locally and globally. According to the results, SO is a heuristic algorithm with strong exploration and exploitation capabilities and can be chosen as an alternative algorithm for large-size continuous optimization problems.
Sosyal Ağlarda Topluluk Tespiti İçin Yeni Bir Algoritma: Ayrık Denge Optimizasyonu
(Konya Teknik Üniversitesi, 2022) Koç, İsmail
Modern ağ bilimi, karmaşık sistemleri anlamlandırmada önemli ilerlemeler sunmaktadır. Gerçek sistemleri temsil eden grafların en alakalı özelliklerinden biri topluluk yapısıdır. Bu tür topluluklar, örneğin insan vücudundaki dokular veya organlar gibi benzer bir rol oynayan, bir grafın oldukça bağımsız bölümleri olarak düşünülebilir. Topluluk tespiti (CD), sistemlerin genellikle graflarla temsil edildiği sosyoloji, biyoloji ve bilgisayar bilimlerinde büyük önem taşımaktadır. CD probleminin araştırılması birçok farklı algoritmayı motive etmesine rağmen, çoğu hesaplama maliyeti nedeniyle büyük ölçekli sosyal ağlar için uygun değildir. Ayrıca, olası topluluk yapılarını tanımlamanın yanı sıra, birçok pratik senaryoda keşfedilen toplulukların nasıl tanımlanacağı ve açıklanacağı da önemlidir. Bu tür gerekçelerde dolayı bu çalışmada klasik yöntemler yerine optimizasyon algoritmasının kullanılması tercih edilmiştir. Optimizasyon algoritması olarak ise son yıllarda geliştirilmiş olan Denge Optimizasyon (EO) Algoritması CD problemine uyarlanmıştır. EO temel versiyonu sürekli problemlerin çözümü üzerine önerildiğinden, ayrık bir problem olan CD problemi için EO yöntemi ayrık hale getirilmiştir. Deneysel çalışmalarda beş farklı sosyal ağ kullanılmıştır. Tüm çalışmalar adil bir kıyaslama yapabilmek için eşit koşullarda gerçekleştirilmiştir. EO algoritması iki farklı makalede yer alan önemli algoritmalarla çözüm kalitesi ve zaman açısından kıyaslanmıştır. Bu sonuçlara göre EO algoritmasının sosyal ağlarda CD probleminin çözümünde çok başarılı olduğu görülmüştür.
Three Different Modified Discrete Versions of Dynamic Arithmetic Optimization Algorithm for Detection of Cohesive Subgroups in Social Networks
(2023) Koç, İsmail
Many networks in nature, society and technology are represented by the level of organization, where groups of nodes form tightly connected units called communities or modules that are only weakly connected to each other. Social networks can be thought of as a group or community, which are groups of nodes with a large number of connections to each other. Identifying these communities by modularity helps to solve the modularity maximization problem. The modularity value determines the quality of the resulting community. Community detection (CD) helps to uncover potential sub-community structures in the network that play a critical role in various research areas. Since CD problems have NP-hard problem structure, it is very difficult to obtain the optimal modularity value with classical methods. Therefore, metaheuristics are frequently preferred in the literature for solving CD problems. In this study, the DAOA algorithm, which has been recently proposed for solving continuous problems, is adapted to the CD problem. In order to improve the solution quality of the DAOA algorithm, some modifications were made in the core parameters. In addition, global and local search supports were added to the DAOA algorithm and three different modifications were applied to the algorithm in total. According to the results performed under equal conditions, among the three modified algorithms, the algorithm with parameter modification was the best in 2 out of 5 networks. DAOA with global search was the best in 3 networks, while the algorithm with local search was the best in 2 networks. However, the basic DAOA could not achieve the best result in any of the 5 networks. This clearly shows the success of the modifications on the algorithm. On the other hand, when compared with the algorithms in the literature, the proposed DAOA algorithm achieved 80% success out of 10 algorithms in total. This shows that the proposed DAOA algorithm can be used as an alternative for discrete problems.
Yeni Bir İkili Sürüş Eğitim Tabanlı Algoritma Üzerinde Transfer Fonksiyonlarının İncelenmesi
(2023) Koç, İsmail
Kapasitesiz Tesis Yerleşim Problemi (UFLP), tesislerin optimal yerleşimini belirleyen NP-zor bir problemdir. UFLP, NP-Zor problem grubundan olduğu için, bu problemlerin büyük örneklerini çözmek için kesin yöntemlerin kullanılması, optimal çözümü elde etmek için gereken yüksek hesaplama süreleri nedeniyle ciddi şekilde sorun teşkil edebilir. Bu çalışmada, problemin karmaşıklığından dolayı sürü zekası algoritması tercih edilmiştir. Son yıllarda sürüş eğitimi ilkelerine dayalı olarak geliştirilen popülasyon tabanlı bir algoritma olan Sürüş eğitim tabanlı (DTBO) algoritması UFLP probleminin çözümünde kullanılmıştır. DTBO’nun temel versiyonu sürekli problemlerin çözümünü ele aldığından söz konusu algoritmanın ikili problemlerin çözümüne uyarlanması gerekmektedir. Bunun için literatürde kullanılan dokuz farklı transfer fonksiyonu yardımıyla DTBO algoritması ikili problemlerin çözümüne uygun olarak tasarlanmıştır. Deneysel çalışmalar transfer fonksiyonlarının adil kıyaslanabilmesi için eşit koşullarda altında gerçekleştirilmiştir. Gerçekleştirilen deneysel çalışmalarda dokuz transfer fonksiyonu içerisinden ikili Mode-DTBO algoritmasının en başarılı algoritma olduğu görülmektedir. Bu sonuçlara göre Mode tabanlı DTBO algoritmasının küçük, orta ve büyük ölçekli tüm problem setlerinde hem çözüm kalitesi açısından hem de zaman açısından çok başarılı olduğu görülmektedir. Ayrıca DTBO algoritması IWO (Yabani Ot Algoritması – Invasive Weed Optimization) algoritmasına ait 3 farklı transfer fonksiyonuyla (Mode, Sigmoid ve Tanh) da kıyaslanmıştır. Karşılaştırmalı sonuçlar incelendiğinde 12 problemin 8’inde (orta ve büyük ölçekli problem) Mode-DTBO yaklaşımının IWO’ya ait 3 farklı yaklaşımın hepsinden çok daha başarılı olduğu görülmüştür. Bununla beraber, küçük boyutlu 4 problem üzerinde ise Mode fonksiyonunu kullanan her iki algoritmanın da optimal değeri yakaladığı görülmüştür. Sonuç olarak, Mode-DTBO yönteminin ikili problemlerin çözümünde çok etkili bir alternatif sunacağı söylenebilir.

Browse

Browsing Bilgisayar ve Bilişim Fakültesi Koleksiyonu by Department "Fakülteler, Mühendislik ve Doğa Bilimleri Fakültesi, Yazılım Mühendisliği Bölümü"