The use of Q-learning methods in multi-robot systems is a challenging area. Multi-robot systems have dynamic and partially observable nature because of robot’s independent decision-making and acting mechanisms. Whereas, Q-learning is defined on Markovian environments theoretically. One way to apply Q-learning in multi robot systems is centralized learning. It learns optimal Q-values for state space of overall system and joint action spaces of all agents. In this case, the system can be considered as stationary and optimal solutions can be converged. But, centralized learning requires full knowledge of the environment, perfect inter-robot communication and good computational power. Especially for large systems, the computational cost becomes huge because of exponentially growing learning space size with the number of robots. The proposed approach in this study, subG-CQL, divides the overall system into small-sized sub-groups without adversely affecting the system's task performing abilities. Each sub-group consists of less number of robots performing less tasks and learns in centralized manner for its own team. So, the learning space dimension is reduced to a reasonable level and required communication remains limited to the robots in the same the sub-group. Due the centralized learning is used, it is expected that the successful results are achieved. Experimental studies show that the proposed algorithm provides increase in the task assignment performance of the system and efficient use of system resources.

Computer assisted diagnosis (CAD) systems have been used frequently in recent years in order to create a doctoral assistance decision support system using various patient information. In this study, it was aimed to compare the success of the Wechsler Intelligence Scale for Children (WISC-R) profiles by the decision trees algorithm applied to the CAD system, and the classification success in the detection of the border intelligence (BI), mild and moderate intellectual disability (ID). The data set of the study was formed by using WISC-R test results of 132 patients (50 BI, 61 mild ID and 21 moderate ID) diagnosed according to DSM-V. In order to compare the effect of WISC-R scores on the outcome, 132 patients' test scores: total, verbal and performance intelligence scores, verbal and performance intelligence subscale scores and 3 separate data sets were formed. The all the score types of the WISCR test, the first two node total intelligence scores were selected, and 108 of 132 patients were classified regardless of another attribute. In the decision tree, the first and third datasets containing the total intelligence section score type achieved a high classification success rate of 0.91. The results of this study show that the total intelligence score of the WISC-R profiles is the most effective in the decision trees algorithm in BI, mild and moderate ID diagnosis, and that CAD systems are possible.

In this study, real-time application of load cell-based Segway has been realized. The control of the system is provided by the PID controller with four load cells placed on the Segway. The analog signals of the weight information coming from the load cell are converted into digital signals with the 24-bit resolution HX711 module. The data converted into digital signals are interpreted through the microprocessor over the synchronous serial communication protocol. The output of the dynamic model of the system can be updated instantly, according to the measured weight information. This process updates the maximum pitch attitude. Thus, the control of the Segway is facilitated and the risk of the user falling off the vehicle is reduced. The control of the system is carried out using an ARM architecture based STM32F103C8T6 microprocessor. Current values and motor rotation speed information of the Brushless Direct Current Motor (BLDC) are obtained during the real time application of the vehicle. In addition, 12 different data are recorded on the SD card with using the SD card module. As a result of the data recorded on the card, the correlation value of each load cell is obtained as 0.99 in the repetition test of the load cells. As a result, the measurement error rate of the weight on the Segway is obtained as 1%.

Device-to-Device (D2D) communication is one of the promising technology for the future 5G networks. Utilizing D2D in cellular networks has advantage in terms of capacity and delay. However, in D2D underlay cellular setting, the main concern is quality of service (QoS) for the cellular user due to the mutual interference between D2D user and the cellular user (CU). To utilize the gain brought by D2D setting without violating QoS of the CU, resource sharing is an important design criteria. To this end, we present an optimization model to investigate a resource sharing problem combined with scheduling in a D2D uplink underlay setting. We have used the proposed model to investigate an example resource sharing scenario, in which multiple D2D pairs share the uplink resource of CU, and identified delay and sum throughput for different parameter settings. We observed that there is a significant gain in terms of sum-throughput in allowing a small number of D2D pairs to re-use the cellular resources.

An image quality metric is proposed by introducing a new framework for full reference image quality assessment from the perspective of image patch manifolds. Assuming that most natural scenes are sampled from low dimensional manifolds or submanifolds, perceived image degradations in structural variations can be quantitatively evaluated on the surfaces of highly nonlinear image manifolds. Manifold distortion image quality index first characterizes intrinsic geometric properties of the locally linear manifold structures of spatially local patch spaces, and then measures the deviation from the original smooth manifold structure to calculate the distortion index. Experimental results demonstrate a strong promise with a comparison to both subjective evaluation and state-of-the-art objective quality assessment methods.

This paper presents a smart machine learning system for classification of hyperspectral wheat data based on deep learning methodology. For this purpose, the performances of AlexNet and VGG16 models were investigated for the classification of hyperspectral wheat samples. In this study, the Support Vector Machine (SVM) and Softmax classifiers were carried out to predict labels of wheat kernels. In order to evaluate the system performance, a new hyperspectral wheat test dataset was constructed using Visible-Near Infrared images (VNIR) including 50 wheat species with 220 images per specimen, as 11000 samples in total. With experiments applied on newly created test dataset, overall approximated accuracy rates of 96.00% and 99.00% determined by linear SVM classifier, in case of fully connected layer (FC6 and FC7) features for AlexNet and VGG16, respectively. From the Softmax predictions, the 92% and 70% of samples were correctly discriminated based on trained VGG16 and AlexNet models, respectively. The obtained superior results show that using a deep Convolutional Neural Networks (CNN) architecture is more efficient by the means of accurate discrimination of wheat species. The proposed deep learning based categorization system promises high accuracy results for the quality analysis, classification and disease detection in food.

The augmented reality technology, in which virtual images are combined with real world objects in real time, has begun to take part in the education of “Generation Z”, 21st century youth. Today, augmented reality applications that contribute to the learning process are also utilized in the field of nursing education. When database research is conducted with the combinations of “nursing education” and “augmented reality” keywords, it is seen that various augmented reality applications are used for improving English proficiency, anatomy knowledge and some of the nursing practices of nursing students. Research results show that students’ learning experiences are improved with the use of augmented reality method thus learning becomes interesting and useful. This review includes the definition of augmented reality technology which increases the reality of environment and improves learning by embodying abstract concepts, its importance and place in nursing education, and the results of various studies on this subject.

Nature is rich with a vast amount of plant and flower species and because of their great diversity; identification of these species requires expertise in the field. Development of an automatic plant identification system can ease this process. In this work, deep Convolutional Neural Networks and Transfer Learning have been utilized in order to develop such an identification system. Images in the database have been collected from other databases and the web and in total it consists of 5,345 flowers and plant images belong to 76 species. 65 of the species are various flower species and 11 of them are other plant species. Data augmentation techniques has been applied in order to increase the number of images in the database and to improve the generalization capacity of the model. For data augmentation, random rotation at four angles, random brightness change in the range of [-0.2, 0.2] and horizontal flip have been applied. Also preprocessing techniques such as center cropping and normalizing have been applied to images before input them to the model. In automatic plant recognition, 0.9971 accuracy achieved on the training set and 0.9897 accuracy achieved on the test set.

One of the promising areas in human computer interaction is the brain computer interfaces and EEG headsets are widely used technology in this domain. In this study, performance comparison of two different low-cost EEG headsets, NeuroSky Mindwave and Emotiv EPOC EEG, in tasks requiring attention and relaxation, and their user experience and usability evaluations were conducted. There were 12 participants who were asked to perform attention tasks that require high cognitive load and relaxation tasks. While the Affect Grid scale and AttrakDiff questionnaire were used to evaluate the user experience, the NASA Task Load Index and System Usability Scale were used to reveal the usability problems of the devices.
When the statistical results were examined, it was observed that the NeuroSky MindWave was more successful than the Emotiv EPOC in relaxation tasks. However, both have similar results in tasks requiring attention. According to the user experience evaluation results, it was observed that the participants felt tired while using both EEG heads, but were still satisfied with the use of the devices. They reported more positive opinions for NeuroSky Mindwave in terms of usability.