59 sonuçlar
Arama Sonuçları
Listeleniyor 1 - 10 / 59
Yayın Author identification of newspaper columns using style and semantic features(Işık Üniversitesi, 2016-05-17) Yıldız, Ergin Doğan; Solak, Ercan; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıThis study has two major purposes : to implement and compare the author classification results of di erent Naive Bayes Classi ers, and to investigate whether merging individual methods will increase classi cation success or not. The subjects of this study were newspaper columnists. Data was collected from well known public newspapers. This study first investigates Numeric, Nominal, Multinominal NBC, and their various merged versions. We then address each method using cross-validation. The results of the experiments show that merging multiple classi cation methods can increase classifi cation success. Additionally, it depends on how well individual classifi cation models are constructed.Yayın Data access layer code generator(Işık Üniversitesi, 2006) Küçükeren, Gökçe; Kuru, Selahattin; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıLayering techniques are commonly used in order to supply the operational requirements of today's complex enterprise applications. The codes developed for the Data Access Layer (DAL), which is the foundation of most layered applications, increases the time of projects' development phases, and the bugs in these manually written codes makes the data access unreliable. In this thesis, a tool is developed to generate the required codes for DAL. The automatically generated codes using this tool, Data Access Layer Code Generator, decreases the time wasted for DAL development and makes the data access reliable due to their uniform structure. Data Access Layer Code Generator is able to build the foundation of the layered architectures with its user-friendly interfaces integrated on the development platform and with the functionalities provided to the developers.Yayın Graph clustering approach to sentiment analysis(Işık Üniversitesi, 2018-01-24) Kanburoğlu, Ali Buğra; Solak, Ercan; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıIn this thesis, we aim at automatically predicting Turkish movie review scores using adjective clustering. We also measured the reliability of the two popular sentiment lexicons. In order to measure the agreement between these sentiment exicons and human judgments, we designed a ranking experiment using pairwise comparisons. Then, we compared these sentiment lexicons and human judgments, and we gave results that show a moderate level of agreement between lexicons and human judgments. Furthermore, we performed adjective clustering task and singleton scoring to automatically assign scores to Turkish movie reviews. Adjective clustering reached an accuracy of 76%, singleton scoring reached an accuracy of 79%.Yayın Visual obstacle detection and avoidance for indoor mobile robots(Işık Üniversitesi, 2012-09-19) İyidir, İbrahim Kamil; Tek, Faik Boray; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıThis study is a part of a joint team effort to transform a small-scale model car into an autonomous moving robot. This transformation includes several routines that are essential and attached together. Integration of various equipment on the model car is the first step of that routine which is shared among different thesis studies conducted at RAVLAB (Robotics and Autonomous Vehicles Laboratory) of Işık University. Hence. The resulting hardware system which is explained in this thesis is mostly the co-produce of RAVLAB team. The integration is followed by implementing the software required to establish control and communication links between different units. During this thesis, the author has developed a multi-threaded main control software to facilitate obstacle avoiding movements of the robot while reading and analyzing sensory inputs. This thesis study mainly focuses on detection of the obstacles with visual information collected from an ordinary color camera. The main obstacle detection algorithm that is proposed in this thesis is adapted from a powerful background subtraction algorithm ViBe [1] (Visual Background Extractor). The proposed algorithm uses the model of the ground plane in order to detect obstacles. A different ground plane model is kept for each pixel location as in ViBe. The proposed algorithm is robust against illumination differences, shadows and the changes in the appearance of the ground plane. A comparison is provided (using ground trurh data) with another obstacle detection algorithm [2] which also uses a ground plane based model to detect obstacles. The results of the proposed algorithm under different conditions compared to a counterpart. In addition to the obstacle detection, during this study two obstacle avoidance algorithms are developed to facilitate navigation of the robot in indoor environments. The experiments show that the robot is able to move while avoiding obstacles by using ultrasonic sensors as well as using the visual camera input.Yayın Data collection and analysis on vanet(Işık Üniversitesi, 2009-02-01) Karatoy, Hilal; Solak, Ercan; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıThe communication between devices is provided by wireless networks and it is growing rapidly, dependent on the needs. Wireless networks strike the attention of people throughout the world, and the ad hoc network is part of this attention. Ad hoc networks do not need structural mechanisms and are mostly used by the mobile nodes. In ad hoc networks the nodes act as a router and communicate with the other nodes. Ad hoc networks can be used in various places such as campus, shopping centers, buildings and vehicles. Ad hoc network systems without any infrastructure have two kinds of nodes which are mobile and fixed. Actually, node mobility is not taken into account. The aim of this thesis is to find the cheapest way to provide the practical statistical data that will be saved after communication between vehicles depend on road constructing, renewing, development and time schedule of the drivers etc. The data used in the analysis were the route and schedule information of the local city buses. Istanbul Metropolitan Municipality provided the coordinates of the bus stops and time tables for the data collection part. We used MATLAB to edit the data in data analysis part. We used 802.11 standards for the communication of the buses and C++ to create the protocol of data transmission. We wrote Google Earth scripts to observe the data correctness and coded TCL scripts for simulation in NS2 to see and analyze the data transfer between nodes, i.e., the buses. Different ranges of communication are used and analyzed, and their results are shown in this project. To collect more data, thousands of nodes are used, which represent more than 40 bus lines and their paths from different continents including Asia and Europe, which constitute the two sides of Istanbul.Yayın Weighted bipartite crossing minimization applications on biclustering and graph unions(Işık Üniversitesi, 2009-02-03) Sözdinler, Melih; Erten, Cesim; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıBiclustering gene expression data is the problem of extracting submatrices of genes and conditions exhibiting significant correlation across both the rows and the columns of a data matrix of expression values. We provide a method, LEB (Localize-and-Extract Biclusters) which reduces the search space in to local neighborhoods within the matrix by first localizing correlated structures. The localization procedure takes its roots from effective use of graph-theoretical methods applied to problems exhibiting a similar structure to that of biblustering. Once interesting structures are localized the search space reduces to small neighborhoods and the biclusters are extracted from these localities. we evaluate the effectiveness of our method with extensive experiments both using artificial and real datasets. Finally, We also used our crossing minimization heuristics for graph visualization in a layered fashion.Yayın An infrastructure and a methodology for business process management(Işık Üniversitesi, 2006-03) Güçlü, Tuncay Kamil; Kuru, Selahattin; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıThis thesis explains in detail an infrastructure and a methodology for business process management as BPM requires much more of a partnership between business and IT. With the proper methodology the business will get the strategy defined so that IT can create the best infrastructure for the situation. The Infrastructure has been created utilizing the 5 step approach defined in the methodology explained in detail further down in the document. The Infrastructure has been includes a network of Microsoft products such as operating systems, applications and tools. The network is a fully operating small domain of servers and client machines. Once the environment was deployed and ready, an application is created using the Visual Studio .net 2003, as a sample implementation. This infrastructre demonstrates how to implement a BPM solution to any organization using the processes defined in this thesis.Yayın Security analysis of coap and dtls protocols for internet of things applications(Işık Üniversitesi, 2019-08-26) Gürkan, Ali Tunca; Tüysüz Erman, Ayşegül; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıInternet of Things is a very fast growing area. Its requirements and related technologies are changing from day to day. In Internet of Things, devices can communicate with each other with different messaging protocols. The latest messaging protocols are well developed, but they are too heavy to be run on devices developed with old technology. Therefore, these devices have to be operated with old-fashioned protocols. This makes devices vulnerable to security risks. CoAP is a newly developed messaging protocol for constrained devices used in Internet of Things applications. The protocol is a variant of HTTP, so it has similar speci cations. CoAP does not have an embedded security mechanism. Therefore, another protocol called DTLS is used on top of it to provide security. DTLS has powerful functions like handshaking and session processes; however, it is weak against DoS attacks. In this study, we develop a security extension for Internet of Things devices using CoAP with DTLS for secure messaging. DTLS applies handshaking process for every received request. The handshaking process is the most time and resource consuming part of the communication. We propose a security extension to prevent unnecessary messaging during handshaking process of an attacker device that sends a lot of unauthenticated requests. When a client sends requests to a server that has the proposed security extension, the server counts unsuccessful handshaking processes for each client. If the count passes a limit of suspicious requests, the security extension on server adds the client's IP address into a banned IPs list. Until the expiration time, the server does not accept any request from the banned IP address. Our proposed security extension is tested in different scenarios to examine the effects on the network. The results of the experiments show that the enhanced security extension decreases delays on the network and it is helpful for communication between authenticated devices.Yayın Navigation of autonomous vehicle in indoor environment(Işık Üniversitesi, 2011) Yılmaz, Hakan; Ekin, Emine; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıThis study investigates transforming a remote controlled vehicle into an au»tonomous vehicle which explores the environment and is able to find its own location in this unknown static indoor environment. It also constructs the map of this environment simultaneously. The study consists of mainly three parts which are (i) the hardware of autonomous vehicle; (ii) while exploring the envi»ronment furnished with landmarks to detect the landmarks, and to control the vehicle so that it can move towards these landmarks and/or it can avoid them; (iii) once having the ability of moving in the environment safely, locating it- self and constructing the map of environment. The hardware of the vehicle has mostly designed and installed by Robotics and Autonomous Vehicles Laboratory (RAVLAB) members as it is a part of one of the RAVLAB's projects . Moving safely in an unknown environment is a requirement for the car to locate itself and to extract the map of the environment, which are known as localization and map- ping problems respectively. While traveling, at every time step, the car estimates its new position by computing the displacement using its instantaneous speed value. The positions of landmarks in its visible area are also estimated. However, real positions of both the car and the landmarks are usually different then the estimated ones because of several factors including sensor noises, car's voltage regulations ete. We have performed a series of experiments in laboratory . The first set of experiments focused on making the vehicle to navigate towards the elosest object. Second set of experiments focused on making the vehicle to stop in the presence of an object in front of the car. Finally, simultaneous localiza»tion and mapping the environment of the car has been tested. The experiments have revealed encouraging results in the sense that the vehicle can traverse the running environment safely. Although, localization and mapping results of the vehicle were not free of error, we were able to get the expected results at the end.Yayın An hybrid approach to solve traveling salesman problem using genetic algorithm(Işık Üniversitesi, 2014) Asmazoğlu, Cengiz; Yıldız, Olcay Taner; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans ProgramıTSP is a challenging and popular problem from combinatorial optimization. TSP is often tackled with well known heuristic genetic algorithm. Due to the nature of the TSP, traditional GA's stay poor when competing with other approaches. Traditional crossover and mutation operators do not satisfy TSP needs. These operators mostly end up with illegal tours. For this reason, researchers proposed many adaptive elements and cooperation of other algorithms. When the logic of GA is combined with these elements, high quality solutions both in time and path length are obtained. In this research, we analyze successful elements from the literature to use them efficiently in a novel algorithm. We also propose a new selection method which works well with our operators. We extend the abilities of greedy crossover and untwist local operator to utilize in our hybrid approach. Multiple populations collaborate together to achieve better solutions. According to the experimental results, proposed novel elements outperforms their counterparts in the TSP literature. Multiple population approach provides better quality solutions.
