Bilgisayar Mühendisliği Bölümü/Makale Koleksiyonu

Bu koleksiyon için kalıcı URI

https://hdl.handle.net/20.500.12395/40822

Listeleniyor 1 - 20 / 44

A new stability approach using probabilistic profile along direction of excavation
(SHAHROOD UNIV TECHNOLOGY, 2020) Turanboy, A.; Ülker, E.; Küçüksütçü, C. Burak.
Estimation of the possible instability that may be encountered in the excavation slope(s) during the planning and application steps of the rock excavation processes is an important issue in geoengineering. In this paper, a modelling method is presented for assessing the probability of wedge failure involving new permanent or temporary slope(s) along the planned excavation direction. The geostructural rock slopes including wedge blocks are determined geometrically in the first step. Here, a structural data analysis system that includes a series of filterings, sortings, and linear equations used to reveal the necessary geometric conditions for the wedge form is developed and used. The second step involves the 3D visualization and Factor of Safety (FS) using the limit equilibrium analysis of wedges on both the actual and planned new excavation surfaces. The last step is the Monte Carlo simulation, which is used in assessing the instabilities on the actual and planned new excavation surfaces. These new slope surfaces that have not yet been excavated are called the virtual structures. As a result of this work, the mean and probabilistic FS variations in the planned excavation direction are obtained as profiles. We suggest the preliminary guidelines for the mean and probability of the wedge failure in the excavation direction. The model is tested on a motorway cut slope. The FS results obtained from the Monte Carlo simulation calculations are compared with the mean results and the changes are revealed with the reasons.
Semi-supervised fuzzy neighborhood preserving analysis for feature extraction in hyperspectral remote sensing images
(SPRINGER LONDON LTD, 2019) Akyürek, Hasan Ali.; Koçer, Barış.
Semi-supervised feature extraction methods are an important focus of interest in data mining and machine learning areas. These methods are improved methods based on learning from a combination of labeled and unlabeled data. In this study, a semi-supervised feature extraction method called as semi-supervised fuzzy neighborhood preserving analysis (SFNPA) is proposed to improve the classification accuracy of hyperspectral remote sensing images. The proposed method combines the principal component analysis (PCA) method, which is an unsupervised feature extraction method, and the supervised fuzzy neighborhood preserving analysis (FNPA) method and increases the classification accuracy by using a limited number of labeled data. Experimental results on four popular hyperspectral remote sensing datasets show that the proposed method significantly improves classification accuracy on hyperspectral remote sensing images compared to the well-known semi-supervised dimension reduction methods.
PSO-based improved multi-flocks migrating birds optimization (IMFMBO) algorithm for solution of discrete problems
(SPRINGER, 2019) Tongur, Vahit.; Ülker, Erkan.
In this paper, we proposed an improved migrating birds optimization algorithm to solve discrete problem. It is a metaheuristic search algorithm that is inspired by V formation during the migration of migratory birds. Proposed algorithm has two main modifications on basic migrating birds algorithm. Firstly, multi-flocks are used instead of single flock in order to avoid local minimum. Secondly, these flocks interact with each other for the more detailed search around flock that has got better solutions. This interaction is inspired by particle swarm optimization algorithm. Also, insertion method is used for neighborhood in migrating birds optimization algorithm. As a discrete problem, traveling salesman problem is chosen. Performance of the proposed algorithm is tested on some of symmetric benchmark problems from TSPLIB. Obtained results show that proposed method is superior to basic migrating birds algorithm.
Optimal placement of wind turbines using novel binary invasive weed optimization
(UNIV OSIJEK, TECH FAC, 2019) Beşkirli, Mehmet.; Koç, Ismail.; Kodaz, Halife.
Wind turbines - which are significant in terms of clean energy production globally - are environmentally friendly, consistent and economical systems. Wind turbines, due to developing technology, have become one of the most widely used renewable energy resources, and every country has worked to satisfy its electricity demands with the help of wind energy. As the importance of wind energy increases all around the world, the importance of wind turbine placement also rises. In this study, the aim was to position wind turbines over a certain area of a wind farm to obtain maximum turbine power with minimum investment cost, thereby achieving the highest power efficiency. The experimental studies were conducted over a 2x2 km area; this area was divided into a 10x10 grid, and a 20x20 grid for more efficient placement. Because these operations occurred in a binary search space, Invasive Weed Optimization (IWO) - normally used to solve unceasing optimization problems - was used in this study by obtaining fourteen different binary Invasive Weed Optimization (BIWO1 to BIWO14) algorithms with the help of ten different transfer functions (four from the sshaped family, four from the v-shaped family, two based on modulo 2, ceil, ceil-round, ceil-floor and round-floor). The proposed method was compared with other studies carried out in the binary search space found in published literature. As a result, it was seen that the proposed algorithm was an efficient algorithm for solving the problem of wind turbine placement to achieve an optimal placement.
Multilevel image thresholding selection based on grey wolf optimizer
(GAZI UNIV, 2018) Koc, Ismail; Baykan, Omer Kaan; Babaoglu, Ismail
Multilevel thresholding is an important image process technique for image processing and pattern recognition. Selecting an optimal threshold value is one of the most crucial phase in image thresholding. While bi-level segmentation contains separating the original image into subdivided sections with help of a threshold value, multilevel segmentation involves multi threshold values. Especially in multilevel image tresholding, the computational time of detailed search increases exponentially with the number of preferred thresholds. For compelling problems, swarm intelligence is known as one of the successful and influential optimization methods. In this paper, the grey wolf optimizer (GWO), a recently proposed swarm-based meta-heuristic which imitates the social leadership and hunting behavior of gray wolves in nature is employed for solving the multilevel image thresholding problem. The experimental results on standard benchmark images indicate that the grey wolf optimizer algorithm is comparable with other state of the art algorithms.
Improved Nelder-Mead Optimization Method in Learning Phase of Artificial Neural Networks
(2018) Koçer, Hasan Erdinç; Merdan, Mustafa; Ibrahım, Mohammed Hussein
Artificial neural networks method is the most important/preferred classification algorithm in machine learning area. The weightson the nets in artificial neural directly affect the classification accuracy of artificial neural networks. Therefore, finding optimum values ofthese weights is a difficult optimization problem. In this study, the Nelder-Mead optimization method has been improved and used fortraining of artificial neural networks. The optimum weights of artificial neural networks are determined in the training stage. Theperformance of the proposed improved Nelder-Mead-Artificial neural networks classification algorithm has been tested on the mostcommon datasets from the UCI machine learning repository. The classification results obtained from the proposed improved Nelder-Mead-Artificial neural networks classification algorithm are compared with the results of the standard Nelder-Mead-Artificial neural networksclassification algorithm. As a result of this comparison, the proposed improved Nelder-Mead-Artificial neural networks classificationalgorithm has given best results in all datasets.
A Modified Artificial Algae Algorithm for Large Scale Global Optimization Problems
(2018) Koçer, Havva Gül; Uymaz, Sait Ali
Optimization technology is used to accelerate decision-making processes and to increase the quality of decision making inmanagement and engineering problems. The development technology has made real world problems large and complex. Many optimizationmethods that proposed for solving large-scale global optimization (LSGO) problems suffer from the “curse of dimensionality”, whichimplies that their performance deteriorates quickly as the dimensionality of the search space increases. Therefore, more efficient and robustalgorithms are needed. When literature on large-scale optimization problems is examined, it is seen that algorithms with effective globalsearch ability have better results. For the purpose, in this paper Modified Artificial Algae Algorithm (MAAA) is proposed by modifyingoriginal version of Artificial Algae Algorithm (AAA) inspiring by Differential Evolution Algorithm (DE)’s mutation strategies. AAA andMAAA are compared with each other by operating with the first 10 benchmark functions of CEC2010 Special Session on Large ScaleGlobal Optimization. The results show that hybridization process that applied by updating an additional fourth dimension with mutationstrategies of DE after the helical motion of the AAA algorithm, contributes exploration phase and improves the AAA performance onLSGO.
Free Cooling Potential of Turkey for Datacenters
(2018) Güğül, Gül Nihal
Electricity consumption for cooling in data centers is increasing rapidly. The heat released from equipment’s in datacenters is dischargedinto the room and room has to be kept at acceptable temperatures. Free cooling technology is more efficient way instead of usingconventional systems in cold regions to reduce the electricity consumption for cooling. It is useful to know the free cooling potential ofa location before installing the free cooling system. In this study, free cooling potential of six cities (Ankara, Antalya, İzmir, Erzurum,Konya and Trabzon) located in different regions of Turkey is investigated according to supply air temperatures from 15 C to 21 C.Then Power Usage Effectiveness (PUE) value of a datacenter with a free cooling chiller is calculated for each city by using the electricityconsumption measurements of cabinets and free cooling system conducted in a datacenter in Ankara, Turkey. Calculations show thatfree cooling system is convenient for Erzurum, Ankara and Konya with PUE values 1.23, 1.37 and 1.37 and the potential calculated tobe almost 100% for eight, six and six months respectively for different supply air temperatures 17 C to 21 C. In addition to coldclimates, free cooling system for Antalya in Tsat19 C is calculated to save 46% energy on annual basis.
Improving an Expert-Supported Dynamic Programming Algorithm and Adaptive-Neuro Fuzzy Inference System for Long-Term Load Forecasting
(2017) Çetinkaya, Nurettin
Load forecasting is very important to manage the electrical power systems. Load forecasting can be analyzed in three different ways as short-term, medium-term and long-term. Long-term load forecasting (LTLF) is inneed to plan and carry on future energy demand and investment such as size of energy plant. LTLF is affected by energy consumption, national incoming per year, rates of civilization, increasing population rates and moreover economical parameters. Some of the forecasting models use mathematical formulas and statistical models such as correlation and regression analysis. In this study, a new effective expert-supported dynamic programming algorithm (ESDP) has been improved. Additionally, adaptive neuro-fuzzy inference system (ANFIS) and mathematical modeling (MM) are used to forecast long term energy demand. ANFIS is one of the famous artificial intelligence and has widely used to solve forecasting problemsin literature. In addition to numerical inputs, ANFIShas linguistics inputs. The results obtained from ESDP, ANFIS and MM are compared to show availability. In order to show error levels mean absolute percentage error (MAPE) and mean absolute error(MAE) are used. The obtained results show that the proposedalgorithms are available.
Training of artificial neural network using metaheuristic algorithm
(2017) Alwaisi, Shaimaa; Baykan, Ömer Kaan
This article clarify enhancing classification accuracy of Artificial Neural Network (ANN) by using metaheuristic optimizationalgorithm. Classification accuracy of ANN depends on the well-designed ANN model. Well-designed ANN model Based on the structure,activation function that are utilized for ANN nodes, and the training algorithm which are used to detect the correct weight for each node.In our paper we are focused on improving the set of synaptic weights by using Shuffled Frog Leaping metaheuristic optimization algorithmwhich are determine the correct weight for each node in ANN model. We used 10 well known datasets from UCI machine learningrepository. In order to investigate the performance of ANN model we used datasets with different properties. These datasets havecategorical, numerical and mixed properties. Then we compared the classification accuracy of proposed method with the classificationaccuracy of back propagation training algorithm. The results showed that the proposed algorithm performed better performance in the mostused datasets.
Transfer Learning in Vehicle Routing Problem for Rapid Adaptation
(ICIC INTERNATIONAL, 2012) Koçer, Barış; Arslan, Ahmet
Vehicle routing problem is a transportation optimization problem, in transporting stuffs from, depot(s) to receivers via vehicle(s) having limited capacity. There are a lot of different problem types in VRP literature with different problem parameters. No VRP method uses past solutions to solve current problems more quickly and to find better solution with less computation. Instead, most of the methods in literature evaluate the changes as a new problem and try to solve the new problem using specialized heuristics. We developed a method which uses past vehicle routes to make new routes quickly for frequently changing conditions, and we achieved good performance improvements over classical methods.
Nurbs Curve Fitting Using Artificial Immune System
(Icic International, 2012) Ülker, Erkan
Non-Uniform Rational B-spline (NURBS) is an industrial standard for Computer Aided Design (CAD) model data representation. For constructing an CAD model from a physical part by curve modeling and dimensional measure, the NURBS design often results in a multi-objective optimization (MOO) problem which cannot be handled as such by traditional single objective optimization algorithms. For large data, this problem needs to be dealt with non-deterministic optimization algorithms achieving global optimum and at the same time getting to the desired solution in an iterative fashion. In order to find a good NURBS model from large number of data, generally the knots, control points and weights are respected as variables. In this paper, the minimization of the fitting error is aimed in order to find a smooth curve and the optimization of the NURBS weights and the knot vector for curve fitting is worked. The heuristic of Artificial Immune System (AIS) was used as a new methodology. The best model was searched among the candidate models by using the Akaike's Information Criteria (AIC). Numerical examples were given in order to show the efficiency of our method.
A Marriage Honey Bee Optimization Approach to the Asymmetric Traveling Salesman Problem
(SILA SCIENCE, 2012) Çelik, Yüksel; Ülker, Erkan
In the travelling salesman problem (TSP), a travelling salesman completes a tour of “n” number of cities by stopping once in each city and completes the tour by returning to his starting point, while minimizing the distance and the cost. The asymmetric travelling salesman problem (ATSP) is the problem in which the cost of travel from city A to B is different from that from B to A. Marriage in Honey Bee Optimisation (MBO) is a meta-heuristic procedure inspired by the mating and insemination process of honey bees. In this study, we seek to use an MBO algorithm for an optimal solution to the ATSP problem, which has previously been solved by different methods. The results of the MBO algorithm for ATSP are compared with Genetic Algorithm (GA), another meta-heuristic method.
Reinforcement Learning Accelerated With Artificial Neural Network for Maze and Search Problems
(IEEE, 2010) Hacıbeyoğlu, Mehmet; Arslan, Ahmet
Reinforcement learning is the problem faced by an agent that must learn behaviour through trial and error interactions with a dynamic environment that lacks the educational examples. Q-learning is one of the most popular algorithms among the reinforcement learning methods. Artificial neural network, as in reinforcement learning, is a sub-entry of machine learning, which can be applied on real frames, the environment of which we do not have sufficient information. Our aim is to enable an autonomous agent placed in a maze to find the shortest path to the target by combining q learning and artificial neural network.
Mineral Identification Using Color Spaces and Artificial Neural Networks
(PERGAMON-ELSEVIER SCIENCE LTD, 2010) Baykan, Nurdan Akhan; Yılmaz, Nihat
Identification of minerals and percentage of their area within a thin section of rock are important for identifying and naming rocks. Colors of minerals are the basic factors for identification. In this study, an artificial neural network is used for the classification of minerals. Optical data of thin sections is acquired from the rotating polarizing microscope stage. For the first analysis we selected a set of parameters based on red, green. blue (RGB) and the second based on hue, saturation, value (HSV) color spaces are extracted from the segmented minerals within each data set. A neural network with k-fold cross validation is trained with manually classified mineral samples based on their pixel values. The most successful artificial network to date is the three-layer feed forward network which uses minimum square error correction. The network uses 6 distinct input parameters to classify 5 different minerals, namely, quartz, muscovite, biotite, chlorite, and opaque. Testing the network with previously unseen mineral samples yielded successful results as high as 81-98%.
Genetic Transfer Learning
(PERGAMON-ELSEVIER SCIENCE LTD, 2010) Koçer, Barış; Arslan, Ahmet
Transfer learning is a method which aims to improve "related" tasks performance. Transfer learning tries to use information gained from related tasks solutions to improve performance of learning strategy. Transfer learning addresses the problem of how to utilize plenty of labeled data in a source domain to solve related but different problems in a target domain, even when the training and testing problems have different distributions or features (Pan, Kwok, & Yang, 2008). In this paper we have used transfer learning to improve performance of genetic algorithms.
Fast Computation of Determination of the Prime Implicants by a Novel Near Minimum Minimization Method
(Tubitak Scientific & Technical Research Council Turkey, 2010) Başçiftçi, Fatih; Kahramanlı, Şirzat
In this study proposed is an off-set-based direct-cover near-minimum minimization method for single-output Boolean functions represented in a sum-of-products form. To obtain the complete set of prime implicants including given on-cube (on-minterm), the proposed method uses off-cubes (off-minterms) expanded by this On-cube. The amount of temporary results produced by this method does not exceed the size of the offset. To make fast computation, we used logic operations instead of standard operations. Expansion off-cubes, commutative absorption operations and intersection operations are realized by logic operations for fast computation. The proposed minimization method is tested on several different kinds of problems and benchmarks results of which are compared with logic minimization program ESPRESSO. The results show that proposed algorithm obtains good results and faster than ESPRESSO.
Energy-Efficient and Fast Data Gathering Protocols for Indoor Wireless Sensor Networks
(Mdpi, 2010) Tümer, Abdullah Erdal; Gündüz, Mesut
Wireless Sensor Networks have become an important technology with numerous potential applications for the interaction of computers and the physical environment in civilian and military areas. In the routing protocols that are specifically designed for the applications used by sensor networks, the limited available power of the sensor nodes has been taken into consideration in order to extend the lifetime of the networks. In this paper, two protocols based on LEACH and called R-EERP and S-EERP with base and threshold values are presented. R-EERP and S-EERP are two efficient energy aware routing protocols that can be used for some critical applications such as detecting dangerous gases (methane, ammonium, carbon monoxide, etc.) in an indoor environment. In R-EERP, sensor nodes are deployed randomly in a field similar to LEACH. In S-EERP, nodes are deployed sequentially in the rooms of the flats of a multi-story building. In both protocols, nodes forming clusters do not change during a cluster change time, only the cluster heads change. Furthermore, an XOR operation is performed on the collected data in order to prevent the sending of the same data sensed by the nodes close to each other. Simulation results show that our proposed protocols are more energy-efficient than the conventional LEACH protocol.
Effects of Principle Component Analysis on Assessment of Coronary Artery Diseases Using Support Vector Machine
(Pergamon-Elsevier Science Ltd, 2010) Babaoğlu, İsmail; Fındık, Oğuz; Bayrak, Mehmet
Artificial intelligence techniques are being effectively used in medical diagnostic support tools to increase the diagnostic accuracy and to provide additional knowledge to medical stuff. Effects of principle component analysis on the assessment of exercise stress test with support vector machine in determination of coronary artery disease are studied in this work. Study dataset consist of 480 patients with 23 features for each patient. By reducing study dataset with principle component analysis method, optimum support vector machine model is found for each reduced dimension. According to the obtained results, optimum support vector machine model in which the dataset is reduced to 18 features with principle component analysis is more accurate than optimum support vector machine model which uses the whole 23 featured dataset. Besides, principle component analysis implementation decreases the training error and the sum of the training and test times.
Effects of Feature Selection Using Binary Particle Swarm Optimization on Wheat Variety Classification
(Springer-Verlag Berlin, 2010) Babalık, Ahmet; Baykan, Ömer Kaan; İşcan, Hazim; Babaoğlu, İsmail; Fındık, Oğuz
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.

Güncel Gönderiler