In this paper, dielectric properties of citrus leaves are predicted with long shortterm memory (LSTM) which is one of the well-known deep neural network (DNN) models and real-time measurements for any moisture content (MC) values in the range of 4.90 to 7.05 GHz at a fixed temperature of 24 C for microwave applications, as a novelty. Firstly, S-parameters of samples are measured with WR-159 waveguide and Waveguide Transmission Line Method. In addition, the MCs of samples depending on their weights are calculated. Thus, the dataset depending on various MC and frequency is obtained with the measurement results to both training and testing the DNN model. Secondly, a total of 4000 datasets are obtained, 80% of which is used for training, and 20% for testing. The proposed DNN model consists of four inputs (f, MC, S 11 , and S 21) and two outputs (ε 0 and ε 00). Finally, the dielectric parameters for the desired MC and f are displayed with the graphical user interface in real-time. Success criteria for the prediction such as mean absolute error, root mean squared error, mean absolute percentage error, and R-squared are calculated. The results indicated that there is good agreement between the measured and predicted ones. R-squared are calculated as 0.962 and 0.968 for ε 0 and ε 00 , respectively.

In the study, it is aimed to classify the apples as rotten and robust by using the deep learning algorithm of the apple images taken from the CAPA database. In the proposed model, the processing steps are image reading, preprocessing and classification of apples, respectively. In the image reading stage, images taken from the image database were used. The applied deep learning architecture consists of introduction, convolutional, activation, pooling, memorization, full connection and conclusion layers. The data used in this architecture are divided into two as 80% training and 20% test data. Four different wavelength, 16 kinds of image combinations were used for the training and testing of the system. At the classification stage, a success rate of 91.25% was achieved in detecting rotten and robust apples. As a result, it is predicted that the proposed model can be used in the fruit processing industry to automatically classify rotten and robust apples.

Yapılan çalışmada günümüzün popüler konularından olan derin öğrenme algoritmaları üzerine bir uygulama geliştirilmiştir. Geliştirilen uygulamada görüntülerden yüz tespiti yapılıp sonrasında görüntüdeki kişinin cinsiyet tahmini gerçekleştirilmiştir. Bu uygulamada Wiki görüntü veri tabanından elde edilen 62328 görüntü kullanılmıştır. Kullanılan görüntüler üzerinde, yüz görüntüsü bulunmayanlar veri setinden çıkartılarak yeni bir veri seti oluşturulmuştur. Oluşturulan veri setindeki görüntülerden, ileri derin öğrenme tekniklerinden biri olan Evrişimsel Sinir Ağları (ESA) yöntemi kullanılarak öznitelikler çıkartılmıştır. Elde edilen öznitelikler Destek Vektör Makinesi (DVM) kullanılarak sınıflandırılmıştır. Sınıflandırma başarısı karmaşıklık matrisi ile gösterilmiş olup, %94,48 başarı oranı ile sınıflandırma işlemi gerçekleştirilmiştir.

No abstract

Purpose The purpose of this paper is to estimate different air–fuel ratio motor shaft speed and fuel flow rates under the performance parameters depending on the indices of combustion efficiency and exhaust emission of the engine, a turboprop multilayer feed forward artificial neural network model. For this purpose, emissions data obtained experimentally from a T56-A-15 turboprop engine under various loads were used. Design/methodology/approach The designed multilayer feed forward neural network models consist of two hidden layers. 75% of the experimental data used was allocated as training, 25% as test data and cross-referenced by the k-fold four value. Fuel flow, rotate per minute and air–fuel ratio data were used for the training of emission index input values on the designed models and EICO, EICO2, EINO2 and EIUHC data were used on the output. In the system trained for combustion efficiency, EICO and EIUHC data were used at the input and fuel combustion efficiency data at the output. Findings Mean square error, normalized mean square error, absolute mean error functions were used to evaluate the error obtained from the system as a result of the test. As a result of modeling the system, absolute mean error values were 0.1473 for CO, 0.0442 for CO2, 0.0369 for UHC, 0.0028 for NO2, success for all exhaust emission data was 0.0266 and 7.6165e-10 for combustion efficiency, respectively. Originality/value This study has been added to the literature T56-A-15 turboprop engine for the current machine learning methods to multilayer feed forward neural network methods, exhaust emission and combustion efficiency index value calculation.