Theses supervised by Prof. Dr. Efendi Nasiboğlu
15 theses · Dokuz Eylül University
Sequential rectangular packing problem in wireless telecommunications with fuzzy extensions
In this thesis, a rectangular packing problem in telecommunications context is considered. Namely, we introduce a resource allocation modeling framework for a sequential two-dimensional packing problem, which may have direct applications in wireless telecommunications area pertaining to the IEEE 802.16 standard. The time dimension implied by the sequential consideration of frames adds a third dimension to the packing problem to some extent. We extend the common features of the frame packing problem to include realistic and state-of-the-art features of the current wireless data transfer processes. Three novel and representative mathematical programming models are developed for the problem, which are intended for contribution both to academic literature and professional practice. The developed models aim optimal usage of the physical layer defined by the standard, which involves data packages sent from a base station to a fixed or mobile user station. The data transmitted for each user are modeled as rectangular blocks, dimensions of which correspond to time duration and frequencies used in data transfer. Placement of these rectangular blocks in a sequence of identical rectangle frames is optimized by the developed models, aiming to maximize profit, minimize waste or minimize the rectangle count. Quality of service constraints such as maximum delay in transfer and minimum data transmission rates restrict the placement of variable-sized rectangles. We present the framework for all models, which handle demand partitioning and rectangle packing simultaneously. The foundations for fuzzy measures and parametrization are also proposed in this thesis, in order to mimic more realistic evaluation of actual network resources for practical problems. Thorough extensive experimentation, the performance of the developed models in terms of both solution times and quality are investigated. We also discuss alternative approaches to improve solution performances for the new models.
Usage of fuzzy logic based data mining methods in analysis of public transportation data
Intelligent Transportation Systems are used to construct and manage public transportation system based on knowledge efficiently and also to increase interest of people for public transport. In scope of this thesis, subtopics of these systems, Advanced Public Transportation Systems and Advanced Traveler Information Systems have been addressed respectively, and two separate applications have been developed. In this study, by examining boarding data obtained from smart cards used in public transportation system in Izmir, estimation of alighting stop for bus mode has been firstly dwelled on. A solution related to these situations, are rarely encountered in literature, about boarding once a day and multiple boarding with same card on same service has been proposed. Moreover comfort degree of passenger in bus has also been estimated by using detailed boarding-alighting information related to requested day, line and its service. In this study, solution has been sought to multi-criteria fuzzy route planning problem as an application of Advanced Traveler Information Systems. For the first time in the literature, fuzzy neighborhood relations between stop-stop, line-stop and line-line, and fuzzy preference degree of stop have been discussed. Information obtained by alighting estimation method has been utilized while stop activity which is one of criteria constituting the fuzzy preference degree of stop has been calculated. It will be possible to develop route planner system similar with human-reasoning thanks to fuzzy concepts used in route planning problem. It is thought that this study will provide contribution to researchers and recently popular issues; smart cities and sustainable mobility.
Veri dağılımının en yakın bulanık gösterimine dayalı zaman serisi etiketlendirmesi
Bu çalışmada üç yeni zaman serileri etiketlendirme yöntemi geliştirildi. Yöntemlerde, danışmansız öğrenme yöntemlerinden biri olan bulanık c-ortalamalar yöntemi zaman serilerinde uygulandı. Her bir küme, merkez değerinin büyüklüğüne göre etiketlendi. Gözlemler küme merkezlerine olan uzaklıklarına göre atandı ve atandığı kümenin etiket değerini aldı. Böylece gözlemlere ait zaman serilerinden, etiketlere ait zaman serileri çıkartılmış olundu. Sonraki adımda, daha düzgün etiket eğrileri elde edebilmek için, -en yakın komşuluk kuralı etiketlere uygulandı. Klasik yöntemden farklı olarak, komşuluklar bulunurken uzaklıklar değil, verinin zaman bazında kendinden önce ve sonra gelen etiketleri dikkate alındı. Önerilen yöntemlerin etkinliği beyin aktiviteleri ile ilgili olan bispektral endeks veri setlerinde araştırıldı ve -en yakın komşuluk kuralının zaman bazında çalıştırılmasının ortalama sınıflama kesinliğinde bir artışa yol açtığı kanıtlandı.Çalışmanın ikinci kısmında, üyelik fonksiyonlarının sınıflama kesinliklerinde artırıcı rolü dikkate alınarak, veri dağılımının en yakın bulanık gösterimleri ile ilgili olan dört yeni teorem geliştirildi. İlgili teoremlerin oluşturulmasında iki yaklaşım kullanıldı. İlk yaklaşımda, veri dağılımının beş noktası parametrik üçgen ve yamuk üyelik fonksiyonlarının beş noktası ile eşleştirildi. İkinci yaklaşımda frekans tabloları kullanıldı. Frekans tablolarındaki normalleştirilmiş yüzdelik değerleri ile sınıf aralıkların orta noktaları dikkate alınarak, amaç fonksiyonları kuruldu. Minimisazyon problemi yardımıyla, verilerin histogramı ile uyumlu parametrik üçgen ve üssel üyelik fonksiyonları elde edildi. Önerilen teoremlerin, sınıflama kesinliğinde artırıcı etkiye sahip olup olmadığını görebilmek için bispektral endeks veri setlerinde sınıflama işlemi yapıldı. Bulunan sınıflama kesinlikleri, literatürde kullanılmış olan başka bir üyelik fonksiyonu yoluyla elde edilenler ile karşılaştırıldı. Veri setlerinin analizi sonucunda, bu tezde geliştirilen üyelik fonksiyon yaklaşımlarının ortalama sınıflama kesinliğinde arttırıcı bir etkiye sahip oldukları kanıtlandı.
Multi expert decision making by using 2 tuple fuzzy linguistic representation and its application to olive oil sensory evaluation
Multi-Expert or Multi-Criteria Decision-Making problems happen frequently in various structures in most areas of daily life. One of the areas, that Multi-Expert Decision-Making problems are used, is the sensory analysis. The evaluation process, performed by panel experts who are experienced in the issue at hand in accordance with some of their senses, in order to gather information on goods, elements, etc., is called Sensory Analysis. Sensory analysis, as it is used in various areas, is widely used in determining the class of Natural Olive Oil. Taking into consideration the importance of the quality of olive oil, which has a special place in Turkey?s agricultural activities, the examination and implementation of sensory analysis, depending on fuzzy linguistic decision analysis. This study aims at implementing the Linguistic Decision Analysis phases, by applying 2-tuple Computational Model on linguistic variables defined in a Multi-granular hierarchical structure in the Multi-Expert Decision-Making Problem. As a result, a linguistic computational model that contributes in observing the senses of tasters for the sensory evaluation of olive oil has been developed, and it has been supported with computer program.
On clustering and classification methods in biosequence analysis
Since human genome studies have brought out a huge number of biosequence data, computational techniques have been developed preventing the vast of cost and time in the management process of these data. In this thesis, new approaches on clustering and classification methods in biosequence ?protein, enzyme sequences? analysis are studied.Classification is a supervised learning algorithm that aims at categorizing or assigning class labels to a pattern set under the supervision of an expert. Therefore, the problem of subcellular location prediction of proteins has been solved by using Optimally Weighted Fuzzy k-NN (OWFKNN). In addition, enzymes have been classified by novel approaches based on minimum-distance classifiers.Clustering is an unsupervised learning technique that aims at decomposing a given set of elements into clusters based on similarity. In this point of view, due to the fact that protein sequences have evolutionary relationship, all protein sequences can be organized in terms of their sequence similarity. A graphical illustration called phylogenetic tree can summarize the relationship between the protein sequences. The construction of phylogenetic tree is based on hierarchical clustering. Thus, we have proposed Ordered Weighted Averaging (OWA) that is most commonly used in multicriteria decision-making, as a linkage method in construction phylogenetic tree. Performance of the OWA-based hierarchical clustering is analyzed by cluster validity indices Root-Mean-Square Standard Deviation (RMSSDT) and R-Squared (RS).
Fuzzy logic and data mining techniques in evaluating of credit risks of companies
In this study, a method that is based on fuzy logic is proposed to determine the credit rate and to be taken assurance amount for the companies which they apply to the bank to get credit and a program is developed for that method. Totaly 109 SMEs (Small and Medium Sized Enterprises ) applied for the credit to the bank which it has branch offices around Turkey and 54 of them are failed in terms of financially, have been examined.At first stage, a model, which classifies as successes and as failed for the credit granting, has been created with supervised education. For that purpose, C&RT decision tree model of SPSS Clementine 10.1 has been used, 33 inputs and 1 output have been evaluated. 8 rules, which establish decision tree mechanism for credit granting, have been found by using C&RT algorithm. Later, fuzzification has been applied to these rules on FIS (Fuzzy Inference System) Editor of Matlab 7.0.1. Mamdani approch has been used based on FIS model, and on created model, 7 inputs and 2 outputs (?to be given credit amount to the company? and ?to be taken assuarance amount from the company?) have been used.Calculated credit results of the proposed method and the results of applied bank credit policy have been compared and it is defined that credit policy of new method is reduced the loss approximately 32% of applied bank credit policy.
Analysis of genetic data via data mining methods and its applications
The prediction of the complete structure of genes is one of the important tasks of bioinformatics, especially in eukaryotes. A crucial part in gene structure prediction is to determine the splice sites in the coding region. Identification of splice sites depends on the precise recognition of the boundaries between exons and introns of a given DNA sequence. This problem can be formulated as a classification of sequence elements into `exon-intron? (EI), `intron-exon? (IE) or `None? (N) boundary classes.In this thesis, we propose a new Weighted Position Specific Scoring Method (WPSSM) to recognize splice sites which uses a position-specific scoring matrix constructed by nucleotide base frequencies. A genetic algorithm is used in order to tune the weight and threshold parameters of the positions on. This method comprises of three phases: learning phase, identification phase and validation phase. In this study, the optimal position weights and threshold parameter are found via genetic algorithm. The proposed WPSS method poses efficient results compared to the performance of various methods proposed in the literature. Computational experiments are conducted on the DNA sequence dataset from `UCI Repository of machine learning databases?.
Data mining on text data and related applications
There is extremely large amount of textual information stored and fast growingly continued to be stored into many storage tools such as database and data warehouse. Thus, reaching needed information is getting slow and hard. Because of this situation, a robust analyzing tool is needed to users. Text mining, which is a branch of data mining, is developed and is still fastly developing tool to handle this problem.Text mining is multidisciplinary that those are ?Natural Language Processing, Information retrieval, Statistics and Data Mining?. In this study, those areas are defined in detail and what parts of those areas are used in text mining.There are many applications that text mining tool is used. In this thesis those are mentioned slightly but automatic text summarization. One of the most used applications in text mining area is automatic text summarization. Needed information has to be reached fastly but after information reached, user must read whole document for interested information. Automatic text summarization task handles this problem and generates a summary of documents to users for time consuming.In this study automatic text summarization task is explained in details and a couple of algorithms are mentioned. Finally, a software coded by using one of those algorithms and then ten Turkish news articles are summarized analyzed by the software.
Fuzzy time series and related applications
Currently, inventing new approaches for modeling the classical time series analysis with last decade?s favorites theme Fuzzy Logic and Sets Theory is going to be popular. In many different scientific models and research areas the Fuzzy Logic Systems are easy to integrate with. Forecasting the short/long distance of future is the main objective of Time Series Analysis and lately it evolves Fuzzy Logic Systems. The main aim in this thesis is evaluating the forecasting or estimation error rate on invented and also improved new models, if they have stronger or weaker affiliations.At the introduction section, effects of Time Series Analysis and Fuzzy Logic Systems in human daily life are separately discussed. The second and third sections include the axioms, definitions of Time Series Analysis and Fuzzy Logic and Sets Theory. The following section after them defines and compares how the newly invented methodology of Fuzzy Time Series gathered. Also the pros and cons of the new system is discussed, so if the forecasting or estimating abilities are superior or not.
On the construction of student groups in a problem based learning system through fuzzy logic considering various objectives
Fuzziness is a concept that was suggested in 1965 by Zadeh that has improved rapidly until today and that has a number of successful applications in many fields. The reason why it has such successful applications and it can be applied in many fields is that it allows expression and analysis of the problems we encounter in daily life more realistically and, thanks to this, it produces more realistic solutions to problems. Therefore, the concept of fuzziness and the theories suggested and the methods developed on this concept are gaining more and more importance day by day.The creation of suitable learning conditions for students is of great importance in the method of problem based learning system which has been continuing in the Department of Statistics at Dokuz Eylül University since 2001. The most important of these conditions is the suitable composition of student groups for the purposes of instruction. For instance, level-based student groups can be composed by dividing students according to their success levels or balanced student groups can be constituted by students of each success level taking place in each group in approximate equal numbers. In addition, student groups can also be constituted by choosing students completely randomly. However, it is quite important that student evaluation grades, which are the fundamental elements used in the group constitution strategies mentioned here, should also be determined suitably. Especially while carrying out such performance evaluations, the opinion formed about the student is both quite difficult to turn into numerical expressions and vary according to each instructor. Thus, there exists the requirement of a system in which the student performance evaluations will be carried out verbally in a more suitable way for human structure of thinking and in which numerical results will later be obtained by using this information.In this dissertation work, a student performance evaluation system and a student group assignment system have been developed by searching for a solution for the above-mentioned problems. Five distinct group assignment strategies have been introduced within the group assignment system. Borland C++ Builder 6.0 Software Development Kit (SDK) was used for the implementation of the mentioned methods with a view to provide a solution.
Construction and analysis of clustering algorithms based on fuzzy relations and their applications to EEG data
In this work, fundamentally two algorithms have been proposed. The first one is the NRFJP (Noise-Robust FJP) algorithm which is a robust version of the known fuzzy neighborhood-based FJP (Fuzzy Joint Points) clustering algorithm. In the NRFJP algorithm each point for which certain eps1 fuzzy neighborhood cardinality is smaller than certain eps2 threshold is perceived as noise. Moreover, in case eps2 is zero, the sensitivity of the NRFJP through noises is turned off, consequently NRFJP Algorithm transforms into FJP algorithm.The second algorithm is the FN-DBSCAN (Fuzzy Neighborhood DBSCAN) algorithm which is a mixture of FJP and density-based DBSCAN (Density Based Spatial Clustering Applications with Noise) algorithms. In the study, the effects of fuzzy neighborhood relation in density-based clustering have been investigated. Besides being a more general algorithm, the FN-DBSCAN algorithm transforms into the DBSCAN algorithm when the crisp neighborhood function is used.The modified version of the FN-DBSCAN algorithm has been developed so as to apply cluster analysis to BIS data. As a result of the computational experiments, it has been observed that FN-DBSCAN based approach gives closer results to the expert?s opinion than the well-known FCM (Fuzzy c-means) clustering algorithm.The codes for the proposed algorithms, NRFJP, FN-DBSCAN and the modified version of FN-DBSCAN to analyze BIS data, have been developed in Borland C++ Builder SDK and they have been designed as an integrated software system.
Beyin - bilgisayar etkileşimi verilerin analizi ve uygulamaları
Vücudun ana kontrol mekanizması olan beyin, anlaşılması zor bir yapıya sahiptir. İnsan beyni üzerinde incelemeler yapılmaya başladığından beri günümüze dek farklı metotlarla, beynin hareketli yapısını anlaşılır hale getirmek için birçok çalışma yapılmıştır. Elektroensefalografi (EEG), bu yöntemler arasında beyin aktivitesini görüntülemeye yarayanlardan bir tanesidir. Beyin – Bilgisayar Etkileşimi, gelişmekte olan ve üzerinde çok fazla sayıda çalışma yapılmamış bir alandır. Yalnızca düşünceler aracılığıyla teknolojik cihazlarla konuşabilme fikri pek çok yeni ufuklar doğurur. Çalışmada kullandığımız Emotiv Epoc+ adlı cihaz, EEG sinyallerini okunabilir hale getirmek için faydalanılan araçlardan bir tanesidir. Kafatasında farklı loblara göre özelleşmiş olarak kendine has bölgelere yerleştirilen 14 farklı elektrot sayesinde bu cihaz aracılığıyla beyinsel aktivite gözlemlenir. Bu tez çalışmasında kullanılan EEG verileri, bahsedilen araç ve yazılım kullanılarak elde edilmiştir. Ayrıca çalışmayı desteklemek amaçlı kullanılan sınıflandırma algoritmalarının testi de UCI veritabanından alınan onaylanmış EEG verileri üzerinde gerçekleştirilmiştir. Beyinde üretilen dalgaların üst üste çakışmasıyla EEG sinyalleri meydana gelir. Bu çalışmada, kullanılan cihaz ve yazılım aracılığıyla bilgisayara aktarılan EEG verileri dosyaya yazdırılmıştır. Verilerin tutulduğu dosyalara gerekli önişleme operasyonları uygulanarak, kullandığımız programlama ortamı olan Visual C# 2015 de veriler ayrıştırılarak okunabilir hale getirilmiştir. Sınıflandırma aşamasında ise K-En Yakın Komşuluk, C x K – En Yakın Komşuluk, Naive Bayesian kullanılmıştır. Sınıflandırma algoritmalarının uzaklık ölçüm yöntemleri içinde ise Öklid, Bray- Curtis, Hellinger ve Cosine benzerlik ölçümleri bulunmaktadır. v Gerçekleştirdiğim tez çalışmasında, aynı denekten temin edilen anlık EEG sinyalleri kullanılmıştır. Deneğe deney sırasında dört temel ana yön (sağ, sol, yukarı, aşağı) gösterilen bir bilgisayar arayüzü sunulmuştur. Bu gösterilen dört yönden fare imlecinin yalnızca birine gitmesinin deneğin zihninde imgelenmesi istenmiştir. Deneğin bu düşünsel süreci EEG ölçüm cihazı yardımıyla kaydedilmiştir. Her bir yön düşüncesi ilgili yön ismiyle etiketlenip dört farklı sınıfa ayrılmıştır. Bu şekilde gruplanan EEG sinyalleri farklı sınıflandırma algoritmaları kullanarak sınıflandırma doğruluk oranlarını karşılaştırmıştır. Nispeten başarılı oran veren algoritma, hangi yönün düşünüldüğünün bilinmediği EEG verileri üzerinde uygulanarak incelenmiştir.
Sosyal medyada sanal zorbalığın tespiti
Siber zorbalık, tüm dünyada olduğu gibi Türkiye'de de büyüyen bir sorundur. Şimdiye kadar elde edilen bulgulara göre, Türkiye'de sosyal medya kullananların siber zorbalığa maruz kalma olasılığı %20'i aşmıştır. Siber zorbalık tespiti İngilizcede çok olmasına rağmen Azerbaycan dili ve Türkçede çok az araştırma bulunmaktadır. Bu sorunu ortadan kaldırmak ve tespit etmek için genellikle makine öğrenimi kullanılmaktadır. Bu çalışmamızda, Azerbaycan dili ve Türkçe metinler üzerinde yapılmış siber zorbalıkları tespit etmek için farklı makine öğrenmesi algoritmaları kullanılmıştır. Çalışmamız, toplam 4400 adet Azerbaycan dili ve Türkçe yazılmış ve sosyal medyadan toplanan cümlelerden oluşan bir veri seti üzerinde makine öğrenimi teknikleri kullanılarak yapılmıştır. Sınıflandırıcıların performansını değerlendirmek için kesinlik (precision), doğruluk (accuracy), duyarlılık (recall) ve F1-skor kullanılmıştır. Çalışmada, kullanılan iki farklı veri setini de ele aldığımızda Türkçe veri setine göre CountVectorizer için %85.98 doğruluk ve %96.94 F1-skor ile Linear SVM modeli en yüksek sonuçlar vermiştir. Yine aynı model ve veri seti ile Tf-IdfVectorizer için en yüksek %85.77 doğruluk ve %97.85 F1-skor sonuçlarına ulaşılmıştır.
Pamuk yetiştiriciliğindeki hastalıkların derin öğrenme yaklaşımı ile tahmin edilmesi
In this thesis, a study on the detection and prediction of cotton diseases, which is a sub-title of environmental factors that are effective in the cultivation of cotton plants, with the help of image processing and deep learning methods is presented. In the first stage, the images of the cotton plant were preprocessed in order to minimize the problems that may be encountered during the application of the preferred deep learning methods. These data obtained as a result of the preprocessing were used as input data for the optimization of the applied deep learning models. With the help of this input data, the hyper-parameters of Convolutional Neural Networks, Long Short-Term Memory Networks and Convolutional Long Short-Term Memory Networks models are decided. In the last phase, the success rates of the predictions made on random images given as input to these optimized models were evaluated. The results obtained as a result of the study were analyzed and compared with the studies in the literature.
Görüntü tanıma ile bir yapay zeka uygulaması
In this thesis, firstly the base program is developed for implementing and testing image processing algorithms. Image loading (from hard drive and webcam), image slicing and mixing functions are implemented. Puzzle solving algorithm based on one side edge detection with candidate adjacent cells is developed, tested and analysed with different images. As a consequence of this, the base of a new algorithm based on grouping jigsaw pieces (like Tetris game pieces) is applied. Width and height limit conditions are added according to the number of pieces in the puzzles. It has been provided to make multiple solutions automatically to increase the solution performance for choosing the optimal solution. On the second stage, pieces with lowest match scores are swapped among themselves for better solutions. The program consists of Tetris Mode and Unrotated Pieces Mode. In the Tetris Mode, pieces with similarity below the threshold are grouped together. These groups are displayed in the Solutions Window on the GUI. In the Unrotated Pieces Mode, the solutions and their scores are displayed here as images. Simulation of pieces, displaying empty/candidate spaces, puzzle boundaries and displaying the solutions in the Solutions Window are carried out at the same time, during the solution process.