Kahya, M.D. | Seneler, C.
Conference Object | 2018 | UBMK 2018 - 3rd International Conference on Computer Science and Engineering , pp.78 - 83
Over the last decade, distributed software development (DSD) become very popular for most of the major companies. According to the effects of globalization, software development methodologies and practices are also impressed. Because of the underlying philosophy of agile, agile teams are paying more attention to communication that is mostly applicable for collocated teams. Most of the companies that apply distributed development, adopted agile software methodology into their software development lifecycle to reduce temporal, geographical and socio-cultural distance challenges. Regarding this, DSD brings along various challenges to b . . .e conducted by the distributed development team and also difficulties to apply agile practices at different locations. In this paper, the findings from a case study on geographical challenges in three small, one medium and one large distributed agile software development projects are presented. The empirical investigation with twelve people was done at one German-based global company that operates more than twenty countries. Finally, geographical distance based challenges were reported that create risks for communication, coordination, and control processes at DSD. © 2018 IEEE
Inal, G. | Kucuk, G.
Conference Object | 2018 | Proceedings - 2018 6th International Symposium on Computing and Networking Workshops, CANDARW 2018 , pp.190 - 195
Today, processors utilize many datapath resources with various sizes. In this study, we focus on single thread microprocessors, and apply machine learning techniques to predict processors' future performance trend by collecting and processing processor statistics. This type of a performance prediction can be useful for many ongoing computer architecture research topics. Today, these studies mostly rely on history-and threshold-based prediction schemes, which collect statistics and decide on new resource configurations depending on the results of those threshold conditions at runtime. The proposed offline training-based machine learn . . .ing methodology is an orthogonal technique, which may further improve the performance of such existing algorithms. We show that our neural network based prediction mechanism achieves around 70% accuracy for predicting performance trend (gain or loss in the near future) of applications. This is a noticeably better result compared to accuracy results obtained by naïve history based prediction models. © 2018 IEEE
Basborekci, E. | Koyuncu, H. | Hamamci, A.
Conference Object | 2018 | 2017 21st National Biomedical Engineering Meeting, BIYOMUT 2017 , pp.190 - 195
Ultrasonography is the method that is routinely applied for screening and classification of hip dysplasia in newborns. Phantoms, that can mimic the acoustic characteristics of human tissues, are needed in ultrasound training or in the calibration of devices. In our study, polylactic acid (PLA) for simulating the bone tissue and gelatin for soft tissue were aimed to create a newborn hip ultrasonography phantom for training purposes. © 2017 IEEE.
Ergün, O.O. | Özturk, B.
Conference Object | 2018 | 26th IEEE Signal Processing and Communications Applications Conference, SIU 2018 , pp.1 - 4
Following recent advances in digital technologies, many data in various domains have been transformed into digital world and shared with millions of users via social media and web technologies. As a result, big amount of data has presented many challenging problems in different fields, e.g internet of things, artificial intelligence. One of application areas is in food domain. Recognition of food category from images, automatic recipe retrieval from internet and analysis and matching of food images with recipes, ingredients, nutrition values bring cooperation of multi disciplines and technologies. In this work, for the first time, s . . .emantical analysis of Turkish Cuisine is held and various information related to food in Turkish Cuisine is structured in a hierarchical ontology model. A new database containing 50 different food categories and related images is constructed and linked with data such as food properties, recipes, etc. As a result, multimodal information retrieval can be achieved faster in a more semantic way. At the same time, food image classification with deep learning methods is performed and faster connection of recognized food category to related semantic data is provided. © 2018 IEEE
Masazade, E. | Kose, A.
Article | 2018 | IEEE Transactions on Signal Processing66 ( 1 ) , pp.86 - 100
In this paper, we study the target tracking problem in a wireless sensor network. A sensor receives a measurement from an energy emitting target and employs binary quantization to the received measurement to generate its decision. A sinusoidal waveform with a certain duration is then used to transmit the sensor decision to the fusion center (FC). All sensor decisions are transmitted to the FC over erroneous wireless channels based on a time division multiple access scheme. We introduce the proportional time allocation (PTA) algorithm where at each time step of tracking, PTA jointly determines the sensors binary quantization threshol . . .ds and their time allocations devoted for the transmissions of binary sensor decisions. Simulation results show that, PTA optimally and dynamically distributes the available transmission time among the sensors near the target so that the decisions of such sensors become less subject to channel errors, and turns off the non-informative sensors located far away from the target. Hence, PTA both saves from the number of sensors transmitting to the FC and provides better estimation performance as compared to ad hoc equal time allocation approaches. © 2017 IEEE
Erdas, A. | Arslan, E. | Ozturkcan, B. | Yildiran, U.
Conference Object | 2018 | 2018 6th International Conference on Control Engineering and Information Technology, CEIT 2018 , pp.86 - 100
The aim of this work is to solve object classification problem using Neural Networks. Two types of neural networks are used and compared, namely, classical Feed Forward Neural Networks and Deep Convolutional Networks. Success of the designed networks is investigated using CIFAR-10 dataset. © 2018 IEEE.
Yigit, E. | Ucak, C.
Conference Object | 2018 | 2017 10th International Conference on Electrical and Electronics Engineering, ELECO 20172018-January , pp.100 - 104
Power transformers are crucial components of electric power systems for continuity of electricity. The reliability of power system depends greatly on the operational performance of the power transformers under different operating conditions. The most important indicator of the operating performance of oil-immersed transformers is the top-oil and the hottest-spot temperature values. Therefore, different thermal models have been developed in the literature to estimate temperature changes of power transformers. In this study, thermal models developed by IEC Loading guide for oil and winding temperatures of power transformers are studie . . .d. To the best of our knowledge, description of IEC models in the literature have some ambiguous points. Thus, ambiguous points have been clarified and some modification to IEC models has been introduced to have better estimates of top-oil and hottest-spot temperatures of power transformers. With these modifications, top-oil and hottest-spot temperatures, which are given for 250MVA ONAF transformer unit in IEC Annex-b, are presented during different load variations as an example. © 2017 EMO (Turkish Chamber of Electrical Enginners)
Yoner, S.I. | Ertas, G. | Akin, A.
Conference Object | 2018 | 2017 21st National Biomedical Engineering Meeting, BIYOMUT 2017 , pp.100 - 104
The most important problem encountered in Functional Near Infrared Spectroscopy is the loss of stability and reliability of the light emitting diode as a result of an on going heat transfer between the living tissue and the light emitting diode. This condition has an unwanted disruptive effect on the radiated light. Within this study, a current source circuit with temperature feedback is developed to reduce these effects to minimum and preliminary experiments are done with the developed circuit. The developed circuit is an op-amp based constant current source which consists of a digital potentiometer, a temperature sensor and a micr . . .ocontroller. Regarding the temperature reading, microcontroller manipulates the value of the digital potentiometer, thereby manipulating the current supplied to the light emittng diode by the current source, instantly. Preliminary studies showed that, the circuit developed for operation between 20°C and 46°C temperature values is much more beneficial when compared to a simple current source. In future, manufacturing the circuit in a modular structure is planned for clinical applications. © 2017 IEEE
Kayahan, I. | Yildiran, U.
Conference Object | 2018 | 2018 6th International Conference on Control Engineering and Information Technology, CEIT 2018 , pp.100 - 104
Trading wind energy in deregulated markets is a challenging task due to uncertainties involved. To cope with this complication, a significant body of work is devoted to the development of day-ahead bidding strategies based on stochastic programming. However, the problem of real-time operation, which can be defined as the management of the system in balancing markets after day-ahead bidding phase is completed, is not studied well in the literature. Motivated by this fact, in the present work, a stochastic model predictive control (SMPC) based real-time operation method is developed for a transmission-constrained joint wind-PHS system . . .. It is assumed that the generation company participates in the day-ahead market and balancing market as a price taker player. Since real- time operation depends on contracts made a priori, day-ahead bidding is also considered as an integral part and modeled as mixed-integer linear programming (MILP) based stochastic program. Main features of the proposed framework, which distinguish it from the previous studies, are the application of an SMPC strategy for real-time operation and inclusion of transmission constraints in bidding and operation phases. © 2018 IEEE
Ozbal, S. | Südor, H.C. | Keskin, A.U.
Conference Object | 2018 | 2018 Medical Technologies National Congress, TIPTEKNO 2018 , pp.100 - 104
This paper analyzes and reports the dynamical behavior of a jerk system used in biomedical modelling applications. This system employs a hyperbolic tangent function as a single source of nonlinearity. The system trajectories, Poincaré maps and Lyapunov exponents and spectrum, as well as Bifurcation diagram are presented to verify the chaotic dissipative behavior of the system, generating a double-scroll chaotic attractor. © 2018 IEEE.
Ozkul, F. | Palaska, Y. | Maşazade, E. | Erol Barkana, D.
Active participation of patients in exercise is an important factor in maintaining rehabilitation programs. In this study, an adaptive algorithm called partially ordered set master (POSM), and a traditional algorithm (increment/decrement one level at a time) that change the difficulty levels of rehabilitation tasks adaptively for each individual have been evaluated. A small working group of 20 healthy subjects are asked to play a fruit- assisted rehabilitation system with the use of these two adaptive methods. The difficulty of the game is dynamically changed according to the performances (score) of the subjects with these two algor . . .ithms. The physiological signals, performance (score), and subjective reports (arousal/valence) are used to evaluate these two algorithms in terms of engaging the subjects. © 2018 IEEE
Yetkin, A.E. | Hamamci, A.
In this study, landmark detectors which are trained on head model obtained using magnetic resonance (MR) images, are applied on the optical camera images of the same subject. 2 different CNN based region proposal networks are employed for this purpose. The problem due to the domain discrepancy between the training and application domains is considered in domain generalization framework. Although, the performance of both networks on their training domain (MRI) were similar, their performance on the target domain (camera images) differ significantly. © 2017 IEEE.