10 results
Search Results
Now showing 1 - 10 of 10
Conference Object Citation - WoS: 1An Undergraduate Curriculum for Deep Learning(Ieee, 2018) Tirkes, Guzin; Ekin, Cansu Cigdem; Sengul, Gokhan; Bostan, Atila; Karakaya, MuratDeep Learning (DL) is an interesting and rapidly developing field of research which has been currently utilized as a part of industry and in many disciplines to address a wide range of problems, from image classification, computer vision, video games, bioinformatics, and handwriting recognition to machine translation. The starting point of this study is the recognition of a big gap between the sector need of specialists in DL technology and the lack of sufficient education provided by the universities. Higher education institutions are the best environment to provide this expertise to the students. However, currently most universities do not provide specifically designed DL courses to their students. Thus, the main objective of this study is to design a novel curriculum including two courses to facilitate teaching and learning of DL topic. The proposed curriculum will enable students to solve real-world problems by applying DL approaches and gain necessary background to adapt their knowledge to more advanced, industry-specific fields.Article Citation - Scopus: 1Analyzing Students' Academic Success in Pre-requisite Course Chains: A Case Study in Turkey(Tempus Publications, 2018) Karakaya, Murat; Eryilmaz, Meltem; Ceyhan, Ulas; Computer EngineeringThere are several principles which have been accepted as approaches to successful curriculum development. In spite of the differences in the proposed sequencing of topics, all approaches basically depend on the pre-requisite chains to implement their educational approach in the curriculum development for specifying the order of the subjects. In this research, two prerequisite chains representing two different curriculum development approaches are taken into consideration in a case study. The first research question considered is whether academic success in a follow-up course is positively related to success attained in the pre-requisite course. The second one is whether or not the selected curriculum development approach for deciding the chains has a significant impact on the academic success relationships between a pre-requisite and its follow-up course. To answer these questions, course data of 441 undergraduate students who graduated from the Atilim University between Fall 2001 and Spring 2015 semesters were collected and analyzed. The results indicate that the succes levels gained in a pre-requisite and its follow-up course are corelated. Moreover, different cirriculum development methods can affect this corelation. Thus, cirriculum developers should consider appropriate approaches to improve student success for deciding chaining courses and their contents.Article Citation - WoS: 1Citation - Scopus: 1Msct: an Efficient Data Collection Heuristic for Wireless Sensor Networks With Limited Sensor Memory Capacity(Ksii-kor Soc internet information, 2015) Karakaya, MuratSensors used in Wireless Sensor Networks (WSN) have mostly limited capacity which affects the performance of their applications. One of the data-gathering methods is to use mobile sinks to visit these sensors so that they can save their limited battery energies from forwarding data packages to static sinks. The main disadvantage of employing mobile sinks is the delay of data collection due to relative low speed of mobile sinks. Since sensors have very limited memory capacities, whenever a mobile sink is too late to visit a sensor, that sensor's memory would be full, which is called a 'memory overflow', and thus, needs to be purged, which causes loss of collected data. In this work, a method is proposed to generate mobile sink tours, such that the number of overflows and the amount of lost data are minimized. Moreover, the proposed method does not need either the sensor locations or sensor memory status in advance. Hence, the overhead stemmed from the information exchange of these requirements are avoided. The proposed method is compared with a previously published heuristic. The simulation experiment results show the success of the proposed method over the rival heuristic with respect to the considered metrics under various parameters.Article Citation - WoS: 3Citation - Scopus: 3Time-Sensitive Ant Colony Optimization To Schedule a Mobile Sink for Data Collection in Wireless Sensor Networks(Old City Publishing inc, 2015) Karakaya, Murat; Computer EngineeringIn Wireless Sensor Networks, sensor nodes are deployed to monitor and record the changes in their surroundings. The collected data in the sensor memories is transferred to a remote central via static or mobile sinks. Because sensors have scarce memory capacity various challenges occur in gathering the data from the environment and transferring them to the remote control. For instance, a sensor's memory might get completely full with the sensed data if the sensor can not transfer them on time. Then, a memory overflow happens which causes all the collected data to be erased to free the memory for future readings. Therefore, when a mobile sink (MS) is employed to collect data from the sensors, the MS has to visit each sensor before any memory overflow takes place. In this paper, we study the design of a mobile sink scheduling algorithm based on the Ant Colony Optimization (ACO) meta-heuristic to address this specific issue. The proposed scheduling algorithm, called Mobile Element Scheduling with Time Sensitive ACO (MES/TSACO), aims to prepare a schedule for a mobile sink to visit sensors such that the number of memory overflow incidents is reduced and the amount of collected data is increased. To test and compare the effectiveness of the MES/TSACO approach, the Minimum Weighted Sum First (MWSF) heuristic is implemented as an alternative solution. The results obtained from the extensive simulation tests show that the MES/TSACO generates schedules with considerably reduced number of overflow incidents and increased amount of collected data compared to the MWSF heuristic.Article Citation - WoS: 47Citation - Scopus: 66Deep Learning Based Fall Detection Using Smartwatches for Healthcare Applications(Elsevier Sci Ltd, 2022) Sengul, Gokhan; Karakaya, Murat; Misra, Sanjay; Abayomi-Alli, Olusola O.; Damasevicius, RobertasWe implement a smart watch-based system to predict fall detection. We differentiate fall detection from four common daily activities: sitting, squatting, running, and walking. Moreover, we separate falling into falling from a chair and falling from a standing position. We develop a mobile application that collects the acceleration and gyroscope sensor data and transfers them to the cloud. In the cloud, we implement a deep learning algorithm to classify the activity according to the given classes. To increase the number of data samples available for training, we use the Bica cubic Hermite interpolation, which allows us to improve the accuracy of the neural network. The 38 statistical data features were calculated using the rolling update approach and used as input to the classifier. For activity classification, we have adopted the bi-directional long short-term memory (BiLSTM) neural network. The results demonstrate that our system can detect falling with an accuracy of 99.59% (using leave-one-activityout cross-validation) and 97.35% (using leave-one-subject-out cross-validation) considering all activities. When considering only binary classification (falling vs. all other activities), perfect accuracy is achieved.Conference Object Citation - WoS: 1Parking Space Occupancy Detection Using Deep Learning Methods(Ieee, 2018) Akinci, Fatih Can; Karakaya, MuratThis paper presents an approach for gathering information about the availabilty of the parking lots using Convoltional Neural Network (CNN) for image processing running on an embedded system. By using an efiicent neural network model, we made it possible to use a very low cost embedded system compared to the ones used in previous works on this topic. This efficient model's performance is compared to one of the models that proved its accuracy in image classification competitions. In these tests, we used datasets that has thousands of different images taken from parking lots in different light and weather conditions.Conference Object An Iot Application for Locating Victims Aftermath of an Earthquake(Ieee, 2017) Karakaya, Murat; Sengul, Gokhan; Gokcay, ErhanThis paper presents an Internet of Things (IoT) framework which is specially designed for assisting the research and rescue operations targeted to collapsed buildings aftermath of an earthquake. In general, an IoT network is used to collect and process data from different sources called things. According to the collected data, an IoT system can actuate different mechanisms to react the environment. In the problem at hand, we exploit the IoT capabilities to collect the data about the victims before the building collapses and when it falls down the collected data is processed to generate useful reports which will direct the search and rescue efforts. The proposed framework is tested by a pilot implementation with some simplifications. The initial results and experiences are promising. During the pilot implementation, we observed some issues which are addressed in the proposed IoT framework properly.Article Citation - WoS: 71Citation - Scopus: 87Efficient Route Planning for an Unmanned Air Vehicle Deployed on a Moving Carrier(Springer, 2016) Savuran, Halil; Karakaya, MuratVehicle routing problem (VRP) is a constrained extension of the well-known traveling salesman problem (TSP). Emerging from the current conceptual trends in operations field, a new constraint to be included to the existing VRP parameters is the depot mobility. A practical example of such a problem is planning a route for an Unmanned air vehicle (UAV) deployed on a mobile platform to visit fixed targets. Furthermore, the range constraint of the UAV becomes another constraint within this sample case as well. In this paper, we define new VRP variants by introducing depot mobility (Mobile Depot VRP: MoDVRP) and extending it with capacity constraint (Capacitated MoDVRP: C-MoDVRP). As a sample use case, we study route planning for a UAV deployed on a moving carrier. To deal with the C-MoDVRP, we propose a Genetic Algorithm that is adapted to satisfy the constraints of depot mobility and range, while maximizing the number of targets visited by the UAV. To examine the success of our approach, we compare the individual performances of our proposed genetic operators with conventional ones and the performance of our overall solution with the Nearest Neighbor and Hill Climbing heuristics, on some well-known TSP benchmark problems, and receive successful results.Conference Object Deep Learning and Current Trends in Machine Learning(Ieee, 2018) Bostan, Atila; Sengul, Gokhan; Tirkes, Guzin; Ekin, Cansu; Karakaya, MuratAcademic interest and commercial attention can be used to identify how much potential a novel technology may have. Since the prospective advantages in it may help solving some problems that are not solved yet or improving the performance of readily available ones. In this study, we have investigated the Web of Science (WOS) indexing service database for the publications on Deep Learning (DL), Machine Learning (ML), Convolutional Neural Networks (CNN), and Image Processing to reveal out the current trend. The figures indicate the strong potential in DL approach especially in image processing domain.Conference Object Citation - WoS: 1Ontology-Supported Enterprise Architecture Analysis(Ieee, 2017) Uysal, Murat Pasa; Karakaya, MuratToday, processing integrated information within and between enterprises is increasingly becoming more and more critical, and so is the implementation and evaluation of an Enterprise Architecture (EA). The review of literature on EA evaluation shows several issues. However, the evaluation of EAs has not attracted sufficient attention, and thus, this research area has not been explored thoroughly yet. We believe that in order to ensure consistency, interoperability and computational inferences among EAs, a complete and holistic approach, rather than monolithic, should be developed. Therefore, in this study, we propose an ontology-supported process model for the evaluation of EAs, and present the implementation details. The main contributions of the present study are the improvements realized in the expressiveness, extensibility, and computable power of EAs, and their evaluation techniques. Although the proposed model requires gathering empirical evidences and investigating applications in concrete cases, the first implications of the proposed model indicates its validity and feasibility, and, hence, the initial results are promising for continuing future studies.
