In this study, a new high-capacity, low-distortion and reversible data hiding method (YKKG) that hides a 24-bit color image in a 24-bit color image is proposed. The proposed method divides the 24-bit image to be hidden into 4-bit pieces and hides each piece by reducing it to the 2-bit hide code. In this way, since the 4-bit piece is reduced to 2-bit, the method performs lossy concealment. 2-bit hiding codes are hidden in 2-byte blocks. In the undo process, the method obtains 4-bit pieces from the 24-bit image hidden from the 24-bit stego image, using 2-bit hiding codes in 2-byte blocks, respectively, and the parts are merged back into reversible 24-bit image. Since the method reduces the size of the data to be halved, it offers twice the capacity compared to traditional LSB methods. When the same amount of data is hidden, the method creates lower distortion in the cover image than traditional LSB methods. Peak signal to noise ratio (PSNR) and structural similarity quality criterion (SSIM), which are frequently used in the literature, were used to measure the image quality of the proposed method. In addition, salt & pepper, gaussian, speckle and poisson attack attacks were used to measure the resistance of the proposed method to visual attacks. The test results showed that the proposed method achieved twice as efficient capacity and higher PSNR and SSIM values than the traditional LSB method and the studies in the literature.

Tartarus is a commonly used benchmark problem for genetic programming. However, it has never been fully explored for its difficulty tuning property. Using the data from a previous study in which we have executed millions of Tartarus instances, we contribute to the literature with an equation to estimate their difficulty. Our approach uses four metrics that are embedded into the equation. These metrics are related to the number of clusters and clusters sizes, the distances of boxes to the edges of the board grid, the number of boxes around the agent, and the minimum number of actions for the agent to reach the largest cluster. The coefficients of these metrics have been fit to the data using the general linear model and a mean residual error of ~0.1 has been achieved. This is the first study that can estimate the difficulty of a Tartarus board without modifying the problem in any way.

Valvular diseases may affect one or more of the cardiac valves, which may need to be replaced or restored for effective treatment. The surgical procedure can be guided by a patient-specific and dynamic model containing information complementary to the 2D/3D static images of the valves. To this end, in this study a novel automated model-free aortic valve segmentation method is presented, and its performance is evaluated against expert annotations over conventional contrast- enhanced ECG-gated multislice CT data of the aortic valve at its closed position. Detailed evaluation of the proposed method in 19 real cases revealed an encouraging performance of 3D region growing over Hessian based approach but also demonstrated the complexity of the problem.

In the literature, it is encountered a vast amount of studies related to the production of controllable wheelchairs for people with disabilities or the prediction of activity thought to be performed. In general, the electroencephalography (EEG) signal is transferred to predetermined classes in these studies. These studies consist of the classification of the EEG signal. However, it has been observed that in the recent years, with the developments in the field of artificial learning, the classification has gone beyond, It can be seen that the visual viewed from the EEG signal can be produced. When the limited number of studies using classical generative adversarial networks (GAN) and autu encoder (AE) approaches are examined, it is seen that visuals from EEG signals can be produced roughly.
The original aspect of this study is that it includes mathematical approaches to increase the visual production capability. Classical GAN architectures use random vector input to provide a variety of images produced. With this approach, it is observed that the visuals produced from the EEG signal are of low quality. In the proposed method, the input is considered as two parts (coded EEG and randomness). Variable auto encoder (VAE) and fourier transform (FT) are used to encode the EEG, while two different approaches are proposed for randomness. The use of this original GAN has enabled higher quality visuals to be produced than EEG signals. In order to understand this quality numerically, pre-trained convolutional neural networks (CNN) was used. As a result of experimental studies, While the performance level of the visuals produced from EEG signals with classical GAN is around 93%, it is seen that this level rises to 95% -100% in the proposed approach.

In this paper, we address the problem of uncertainty management in identification of paraphrase sentence pairs. Paraphrase sentences are simply sets/pairs of sentences that express the same facts and/or opinions using different words or order of words. We propose the use of certainty factor (CF) model in paraphrase detection. A set of succeeding paraphrase detection features (generic and distance based features) is built by filtering and this set is used as evidences in CF model. The CF model is evaluated by F1 and accuracy measures on Microsoft Research Paraphrase corpus. The results are compared to the well-known Bayesian reasoning. The experimental results showed that CF model is an alternating paraphrase detection method to Bayes model.

Parkinson's disease is a common neurodegenerative neurological disorder, which affects the patient's quality of life, has significant social and economic effects, and is difficult to diagnose early due to the gradual appearance of symptoms. Examining the discussion of Parkinson’s disease in social media platforms such as Twitter provides a platform where patients communicate each other in both diagnosis and treatment stage of the Parkinson’s disease. The purpose of this work is to evaluate and compare the sentiment analysis of people about Parkinson's disease by using deep learning and word embedding models. To the best of our knowledge, this is the very first study to analyze Parkinson's disease through social media by using word embedding models and deep learning algorithms. In this study, Word2Vec, GloVe, and FastText as word embedding models and Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Long Short Term Memory Networks (LSTMs) as deep learning techniques are blended and used for classification purpose. Extensive experiments are conducted to analyze the sentiments of user comments about Parkinson's disease using word embedding models and deep learning algorithms on English Twitter dataset. The remarkable classification success with 75.12% of accuracy is observed in the experiments through the result of blending Word2Vec as a word embedding model and CNN as a deep learning technique. This study demonstrates the effectiveness of using word embedding models and deep learning algorithms to understand patients' needs, and provides a valuable contribution to the treatment process by analyzing the feelings of Parkinson's patients and their relatives through social media.

Globalization and rapid developments in science and technology lead to an increase in competition and diffraction in the objectives in the production methods. In order to meet the rapidly changing and differentiated needs, manufacturing businesses are left against technological renewal. Especially usage of the data that is collected in electronic media and the ease of access to information forces businesses to review computer systems on point of production management.
Visualization of the data analyzed in the databases is a suitable solution in the decision-making processes of the manufacturing companies. In this context, the dashboard is seen as a good support tool especially for the manufacturing businesses, at a fast and accurate decision-making point.
This article represents a new model approach to accumulated analysis and its sharing for the manufacturing businesses by using the artificial immune system and data mining techniques under the title of the dashboard. In the model, data is increased and handled with clonal selection algorithm. In the analysis stage, the data is clustered with k-means algorithm. The data are visualized by calculating the weighted average and the performance indicators. The visuals that have been obtained will be shared with an app which supports the decision makers with the dashboard rules. Our approach provides a new approaching model to unite, analyze and visualize the collections of data.

Wireless sensor networks are utilized in many different areas such as health, agriculture, security, and entertainment. Since the nodes that constitute wireless sensor networks have limited energy resources, many studies have been carried out on use of their resources in an energy efficient manner. These studies generally focused on duty cycle techniques or constructing energy efficient communication backbones. In wireless sensor systems, the connected dominating sets are generally considered to be used as a backbone. In addition, the deployment of the nodes that harvest their own energy in wireless sensor networks has also been considered in recent studies. In this study, a new distributed algorithm is proposed to construct reusable connected dominating set for wireless networks that possess energy harvester nodes and ordinary nodes which could not harvest its energy. Whenever an energy depletion problem in a node or in more than one node occurs, after a specific interval, the proposed algorithm is re-employed on the alive nodes unless wireless sensor network is disconnected. The proposed algorithm was implemented on SensEH simulation environment based on Cooja which is one of the most commonly used tools in the wireless sensor network area. The new algorithm and the rival algorithm in the area are analyzed with respect to the lifetime of the systems together with the time and the energy consumptions required by the algorithms. According to the results, it is observed that the proposed algorithm can double the total lifetime compared to the rival algorithm.

In this study, the problem is to find a route for a UAV that takes off from Istanbul to observe the damages that may occur after the possible Istanbul earthquake within the first 24 hours. In the problem, 230 candidate grid points that UAV can visit on Istanbul are determined and the weight values combining the risk values based on earthquake degree zones and the population densities of the grid points are calculated for each candidate point. It is aimed to find a route for the UAV to maximize the total weights of the visited grid points under the UAV range constraint. The described problem is adapted to the Orienteering Problem in the literature. Since the Orienteering Problem is an NP-hard problem, a problem-specific genetic algorithm and a simulated annealing algorithm are developed to solve the problem. The parameters of the algorithms are tuned by experiments. 15 different scenarios including the daily number of visits (of taken images) and the airports that the UAV takes and lands off after the earthquake are created and tried to be solved exactly via ILOG and approximately via developed metaheuristics. While the optimal solutions are found for 2 of 15 scenarios via ILOG, the designed genetic algorithm has better solutions and can solve the problem within acceptable CPU times for the rest of the scenarios.

In a power system, the calculation of the voltage amplitudes, phase angle values and transmission losses of buses using the buses with known voltage amplitudes and power values is called power flow problem or load flow problem. Increasing consumption day by day and connecting new energy sources to the power system make the power flow problem more complicated. The power flow problem is great importance for both the electricity production and transmission because the planning the loads that can be connected to the system in the future and usage of the existing transmission lines with full capacity are based on the solution of this problem. Traditional numerical solutions have been used in the solution of this non-linear problem originating from the nature of the system. However, the optimization techniques and search algorithms developed later show that better results can be obtained in solving the power flow problem. In this study, Artificial Bee Colony (ABC), Gray Wolf (GWO) and Particle Swarm Optimization (PSO) algorithms have been applied to the power system, which has the IEEE-14 bus parameters, to optimize the power flow problem using Matlab software. In the conclusion of the study, the voltage amplitude, phase angle, active power loss values of the model power system and calculation times of the algorithms are compared.

The purpose of the study is to develope a high-performance speed sensorless indirect field oriented control for induction motors (IMs). The proposed method is a novel Model Reference Adaptive System (MRAS) and needs only steady-state stator q-axis voltage equation to estimate rotor speed. And also, the proposed speed estimator algorithm removes the voltage transducer requirement in calculations since the algorithm compares the current requlator PI controller output with the calculated q-axis stator voltage. So the system does not need a reference loop since the calculated voltage I adaptive sub-model is directly compared with controller output. This simple equation does not require any rotor parameter and this makes the system immune to the variation of rotor parameters. Moreover, this unique calculation eliminates the requirement of flux estimation thus, the method is less sensitive to pure integration problems. This makes the estimator quite accurate at very low and zero speeds. Moreover, the suggested MRAS technique eliminates the voltage transducer measurement noises so, the low speed accuracy of the speed estimator is increased. Which are validated in simulations using MATLAB/SIMULINK.

Unmanned aerial vehicles (UAVs) are widely used in many fields, both civilian and military. In particular, mostly mini UAVs are used in civilian applications which are preferred both in terms of affordability and availability. However, these vehicles are insufficient for some applications when they are used alone. This inadequacy is often observed in mini-UAVs having limited capacity in terms of energy storage and payloads. As a solution to this challenge, swarms of networked mini UAVs have been proposed. Thus, the UAVs in such networks are assigned with different tasks to provide a solution or this inadequacy. While these networks have many advantages, they also come with challenges. These challenges include limited on-board energy storage, low power and lossy wireless communication interface, and limited useful payload carrying capability. In this study, a new adaptive, balanced and energy efficient clustering mechanism has been proposed for such UAV networks.

The electromagnetic frequency spectrum is divided into different sub-frequency bands. These sub-frequency bands are allocated for different applications. In these days, devices operating in multiple sub-frequency bands provide significant advantages. Devices require antenna structures to operate in multiple frequency bands. Microstrip antennas have become prominent antenna structures with their small size, portable structures and easy integration into other systems. In this study, microstrip antenna structure which can work in multi frequency bands is designed. At the same time, it was used with deep learning methods in optimization of antenna sizes to ensure the optimization of the designed antenna in a shorter time. The operating frequencies of designed antenna structure work in the C and X band as seen in the obtained results. According to IEEE standards, C band is determined between 4 GHz and 8 GHz; X band determined as in 8 GHz and 12 GHz frequency range. In the proposed antenna structure, the ability to operate in multi-band structures was achieved by means of a C-shaped antenna array. In the deep learning methods that will be used in the optimization process, five different Long Short Term Memory (LSTM) models are used. The most important advantage of deep learning methods is that it can achieve satisfactory results by identifying the necessary features for solving difficult and time consuming problems with its own learning ability. In this context, 52 pieces of antenna data were produced. 40 pieces of data were used in the training process and 12 pieces of data were used in the test stage. The lowest root mean square error (RMSE) performance obtained in the test data was determined as LSTM-1 + Dropout layer-1 + LSTM -2 + Dropout layer-2 and 1,0161 error value. The obtained results by proposed method were evaluated in High Frequency Simulation Software (HFSS) program. In experimental results, it was observed that the results produced by the deep learning model and the test data were very close to each other.

In this study, a Julia programming language based system description language and simulation environment that enables fast and effective system simulations together with online and offline data analysis is introduced. In the simulation environment developed, it is possible to simulate discrete time or continuous time, static or dynamical systems. In particular, it is possible to simulate dynamical systems modeled by different types of equations, such as the ordinary differential, random ordinary differential, stochastic differential, differential-algebraic, delayed differential equations, and discrete-time difference equations. During the simulation, the data flowing through the links of the model can be processed online and offline, and specialized analysis can be performed. These analyzes can also be enriched with plugins that can be easily defined using the standard Julia library or various Julia packages. The simulation is performed by evolving the model components individually and parallelly between sampling time intervals. The independent evolution of the components allows the simulation of the models consisting of the components represented by different mathematical equations, while the parallel evolution of components increases the simulation speed.