Elektrik-Elektronik Mühendisliği Bölümü Koleksiyonu

Bu koleksiyon için kalıcı URI

https://hdl.handle.net/20.500.14034/36

Listeleniyor 1 - 20 / 35

Serverless federated learning: Decentralized spectrum sensing in heterogeneous networks
(Elsevier, 2025) Catak, Ferhat Ozgur; Kuzlu, Murat; Dalveren, Yaser; Ozdemir, Gokcen
Federated learning (FL) has gained more popularity due to the increasing demand for robust and efficient mechanisms to ensure data privacy and security during collaborative model training in the concept of artificial intelligence/machine learning (AI/ML). This study proposes an advanced version of FL without the central server, called a serverless or decentralized federated learning framework, to address the challenge of cooperative spectrum sensing in non-independent and identically distributed (non-IID) environments. The framework leverages local model aggregation at neighboring nodes to improve robustness, privacy, and generalizability. The system incorporates weighted aggregation based on distributional similarity between local datasets using Wasserstein distance. The results demonstrate that the proposed serverless federated learning framework offers a satisfactory performance in terms of accuracy and resilience.
A simplified method based on RSSI fingerprinting for IoT device localization in smart cities
(Ieee-Inst Electrical Electronics Engineers Inc, 2024) Dogan, Deren; Dalveren, Yaser; Kara, Ali; Derawi, Mohammad
The Internet of Things (IoT) has significantly improved location-based services in smart cities, such as automated public transportation and traffic management. Estimating the location of connected devices is a critical problem. Low Power Wide Area Network (LPWAN) technologies are used for localization due to their low power consumption and long communication range. Recent advances in Machine Learning have made Received Signal Strength Indicator (RSSI) fingerprinting with LPWAN technologies effective. However, this requires a connection between devices and gateways or base stations, which can increase network deployment, maintenance, and installation costs. This study proposes a cost-effective RSSI fingerprinting solution using IQRF technology for IoT device localization. The region of interest is divided into grids to provide training locations, and measurements are conducted to create a training dataset containing RSSI fingerprints. Pattern matching is performed to localize the device by comparing the fingerprint of the end device with the fingerprints in the created database. To evaluate the efficiency of the proposed solution, measurements were conducted in a short-range local area ( $80\times 30$ m) at 868 MHz. In the measurements, four IQRF nodes were utilized to receive the RSSIs from a transmitting IQRF node. The performances of well-known ML classifiers on the created dataset are then comparatively assessed in terms of test accuracy, prediction speed, and training time. According to the results, the Bagged Trees classifier demonstrated the highest accuracy with 96.87%. However, with an accuracy of 95.69%, the Weighted k-NN could also be a reasonable option for real-world implementations due to its faster prediction speed (37615 obs/s) and lower training time (28.1 s). To the best of the authors' knowledge, this is the first attempt to explore the feasibility of the IQRF networks to develop a RSSI fingerprinting-based IoT device localization in the literature. The promising results suggest that the proposed method could be used as a low-cost alternative for IoT device localization in short-range location-based smart city applications.
Advancing mmWave Altimetry for Unmanned Aerial Systems: A Signal Processing Framework for Optimized Waveform Design
(MDPI, 2024) Awan, Maaz Ali; Dalveren, Yaser; Kara, Ali; Derawi, Mohammad
This research advances millimeter-wave (mmWave) altimetry for unmanned aerial systems (UASs) by optimizing performance metrics within the constraints of inexpensive automotive radars. Leveraging the software-defined architecture, this study encompasses the intricacies of frequency modulated continuous waveform (FMCW) design for three distinct stages of UAS flight: cruise, landing approach, and touchdown within a signal processing framework. Angle of arrival (AoA) estimation, traditionally employed in terrain mapping applications, is largely unexplored for UAS radar altimeters (RAs). Time-division multiplexing multiple input-multiple output (TDM-MIMO) is an efficient method for enhancing angular resolution without compromising the size, weight, and power (SWaP) characteristics. Accordingly, this work argues the potential of AoA estimation using TDM-MIMO to augment situational awareness in challenging landing scenarios. To this end, two corner cases comprising landing a small-sized drone on a platform in the middle of a water body are included. Likewise, for the touchdown stage, an improvised rendition of zoom fast Fourier transform (ZFFT) is investigated to achieve millimeter (mm)-level range accuracy. Aptly, it is proposed that a mm-level accurate RA may be exploited as a software redundancy for the critical weight-on-wheels (WoW) system in fixed-wing commercial UASs. Each stage is simulated as a radar scenario using the specifications of automotive radar operating in the 77-81 GHz band to optimize waveform design, setting the stage for field verification. This article addresses challenges arising from radial velocity due to UAS descent rates and terrain variation through theoretical and mathematical approaches for characterization and mandatory compensation. While constant false alarm rate (CFAR) algorithms have been reported for ground detection, a comparison of their variants within the scope UAS altimetry is limited. This study appraises popular CFAR variants to achieve optimized ground detection performance. The authors advocate for dedicated minimum operational performance standards (MOPS) for UAS RAs. Lastly, this body of work identifies potential challenges, proposes solutions, and outlines future research directions.
Real-time radar classification based on software-defined radio platforms: enhancing processing speed and accuracy with graphics processing unit acceleration
(MDPI, 2024) Oncu, Seckin; Karakaya, Mehmet; Dalveren, Yaser; Kara, Ali; Derawi, Mohammad
This paper presents a comprehensive evaluation of real-time radar classification using software-defined radio (SDR) platforms. The transition from analog to digital technologies, facilitated by SDR, has revolutionized radio systems, offering unprecedented flexibility and reconfigurability through software-based operations. This advancement complements the role of radar signal parameters, encapsulated in the pulse description words (PDWs), which play a pivotal role in electronic support measure (ESM) systems, enabling the detection and classification of threat radars. This study proposes an SDR-based radar classification system that achieves real-time operation with enhanced processing speed. Employing the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm as a robust classifier, the system harnesses Graphical Processing Unit (GPU) parallelization for efficient radio frequency (RF) parameter extraction. The experimental results highlight the efficiency of this approach, demonstrating a notable improvement in processing speed while operating at a sampling rate of up to 200 MSps and achieving an accuracy of 89.7% for real-time radar classification.
The fast and reliable detection of multiple narrowband FH signals: a practical framework
(MDPI, 2024) Aydin, Mutlu; Dalveren, Yaser; Kara, Ali; Derawi, Mohammad
Frequency hopping (FH) is a well-known technique that is commonly used in communication systems owing to its many advantages, including its strong anti-jamming capability. In this technique, basically, radio signals are transmitted by switching the carrier between different frequency channels. As a result, the FH signal is not stationary; hence, its spectrum is expected to change over time. Therefore, the task of detection and parameter estimation of FH signals is very challenging in practice. To address this challenge, the study presented in this article proposes a method that detects and estimates the parameters of multiple narrowband FH signals. In the proposed method, first, short-time Fourier transform (STFT) is utilized to analyze FH signals, and a practical binarization process based on thresholding is used to detect FH signals. Then, a new algorithm is proposed to ensure that the center frequencies of the detected signals are successfully separated. Next, another algorithm is proposed to estimate the parameters of the detected signals. After estimating the parameters for the entire spectrum, an approach is used to detect FH signals. Lastly, the hop-clustering process is applied to separate the hops into groups without time overlap. The simulation results show that the proposed method can be an efficient way for the fast and accurate parameter estimation and detection of multiple narrowband FH signals.
Estimation of weibull probability distribution parameters with optimization algorithms and Foca wind data application
(Gazi Univ, 2024) Köse, Bayram; Işıklı, İbrahim; Sağbaş, Mehmet
In this study, the scale and shape parameters of the Weibull probability distribution function (W.pdf) used in determining the profitability of wind energy projects are estimated using optimization algorithms and the moment method. These parameters are then used to estimate the wind energy potential (WEP) in Foca region of Izmir in Turkey. The values of Weibull parameters obtained using Particle Swarm Optimization (PSO), Sine Cosine Algorithm (SCA), Social Group Optimization (SGO), and Bat Algorithm (BA) were compared with the estimation results of the Moment Method (MM) as reference. Root mean square error (RMSE) and chi-square (chi<^>2) tests were used to compare the parameter estimation methods. The wind speed measurement values of the observation station in Foca were used. As a result of Foca speed data analysis, the annual average wind speed was determined as 6.15 m/s, and the dominant wind direction was found as northeast. Wind speed frequency distributions were compared with the measurement results and calculated with the estimated parameters. When RMSE and chi<^>2 criteria are evaluated together; it can be concluded that each used method behaves similarly for the given parameter estimation problem, with minor variations. As a result, it has been found that the optimization parameters produce very good results in wind speed distribution and potential calculations.
Comparative analysis of mathematics, statistics and physics based algorithms for obtaining optimum weibull probability distribution parameters for power density estimation in wind energy : Loras and Foca examples
(Turkish Soc Thermal Sciences Technology, 2024) Köse, Bayram; Telkenaroğlu, Bekir Can; Demirtürk, Bahar
In this work, the k and c parameters of the Weibull probability distribution function, which is generally used in the feasibility and efficiency studies of wind energy and preferred in electrical energy production, were estimated by Simulated Annealing Algorithm (SA) and Generalized Reduced Gradient Algorithm (GRG). AFunction parameters were also estimated by classical numerical methods, Least Squares Method(LMS), AJustus Empirical Moment Method(EMJ) and Lysen Empirical Moment Method(EML). When comparing the results, the coefficient of determination, the root mean square error (RMSE) and the chi-square distribution criteria(x(2)) Awere used. Wind speed frequency distributions were calculated with the estimated shape and scale parameter and compared with the measurement results. Consequently, better results can be seen from GRG algorithm than the classical numerical methods with coefficient of value of 0.0182 RMSE, determination of 0.8473, and the value x(2) of 0.0079 for Loras and with coefficient of value of 0.0066 RMSE, determination of 0.9793, and the value of 0.0011 for
Single VDCC based memcapacitor emulator circuit without using passive elements and analog multiplier
(Natl Inst Science Communication-Niscair, 2024) Korkmaz, Muhammet Oguz; Sağbaş, Mehmet; Babacan, Yunus; Yesil, Abdullah
In this work, a grounded charge-controlled memcapacitor emulator circuit based on a Voltage Difference Current Conveyor (VDCC) is presented. The proposed circuit is constructed with only one VDCC element and four MOSFETs. It has a transistor-based structure without the use of any passive components. The absence of any passive components makes the circuit eliminating the need for complicated analog components. The suggested circuit does away with the necessity for a mutator, which eliminates the need for an additional separate memristor emulator. The analog multiplier circuit is also not used. Additionally, the designed memcapacitor circuit allows for independent electronic control of both the fixed and variable components, adding to its flexibility and adaptability. To demonstrate the accuracy of the suggested circuit, a SPICE simulation was run using a VDCC constructed with 0.18 mu m TSMC CMOS transistors.
Analysis of C-shaped compact microstrip antennas using deep neural networks optimized by Manta Ray foraging optimization with Lévy-Flight Mechanism
(2021) Biçer, Mustafa Berkan
In recent years, microstrip antennas have become a popular research subject with the increasing use of mobile technologies. With the development of neural networks, the design and analysis of microstrip antennas are carried out quickly with high accuracy. However, optimizing the weight matrices and bias vectors of deep neural learning models is an important challenge for engineering problems. This study presents a deep neural network-based (DNN-based) neural model to estimate the gain and scattering parameter (S11) of C-shaped compact microstrip antennas (CCMAs). For this purpose, the S11 and gain values of 324 CCMAs with different physical and electrical properties were obtained using full-wave electromagnetic simulation software based on the finite integration technique (FIT). The data related to 324 CCMAs were used for the training and testing process. The improved manta ray foraging optimization (MRFO) algorithm based on the Lévy-flight (LF) mechanism was used to optimize the connection weights matrices and bias vectors. The MRFO-optimized model has estimation success for training and testing data as 0.925 and 0.922, in terms of R2 score, respectively. The estimated resonant frequencies using the trained model are compared with the studies in the literature, and an average percentage error (APE) of 0.933% is obtained.
İzmir Menemen Bölgesinin güneş ışınım tahmini için angström-prescott modeli temelli yaklaşımların değerlendirilmesi
(2022) Yıldırım, Hatice Başak
Öz Güneş enerjisi, sınırlı bir ömre sahip olan ve çevreye zarar veren fosil yakıtların yerini alacak önemli bir role ve potansiyele sahiptir. Güneş enerjili sistemleri uygun şekilde tasarlamak, projelendirmek, inşa etmek ve işletmek için güneş ışınımı bilgisi gereklidir. Güneş ışınımı tüm dünyada nadiren ölçülmekle birlikte, diğer meteorolojik parametreler hemen hemen tüm istasyonlarda ölçülmektedir. Güneş ışınımının ölçüldüğü istasyonların veri tabanlarında da boşluklar bulunmaktadır. Bu noktada güneş ışınımının tahmini önem kazanmaktadır. Literatürde güneş ışınımının güneşlenme süresine bağlı olarak hesaplandığı Angström-Prescott modeli ve bu modelin geliştirilmesi yaklaşımıyla oluşturulmuş farklı modeller bulunmaktadır. Bu çalışmada, Menemen ilçesinde küresel güneş radyasyonunu tahmin etmek için yıllık ve mevsimlik modeller oluşturulmuştur. Ayrıca Angström-Prescott yöntemi temelli modellerin güneş ışınım şiddeti tahmini amacıyla kullanımında hangi modelin daha etkin olduğu Menemen ilçesi için araştırılarak karşılaştırılmalar gösterilmiştir. Bu modelleri oluşturmak için MATLAB programı kullanılmıştır. Son olarak bu modeller ile hesaplanan global güneş ışınımı değerleri ve yer ölçümleri R2, SSE, RMSE gibi istatistiksel testler ve hata analizleri ile karşılaştırılmış, hesaplanan ve ölçülen değerler grafiklerle gösterilmiştir.
Frequency analysis of rounded shaped inductive metallic objects in waveguides via some PEC approximations and GSM method
(2022) Aydoğan, Ahmet
A hybrid method is proposed for the frequency analysis of rounded metallic objects inductively loaded in rectangular waveguides. The proposed method combines the efficiency of the generalized scattering matrix method (GSM) and the flexibility of the method of moments (MoM) and the fact that fields cannot exist inside perfect electric conductors. Metallic discontinuities are modelled as a dielectric medium with extreme conductivity and the volume is emptied except the surrounding area. The proposed method is tested against several structures including a band-pass filter composed of metallic rods and an arbitrarily shaped discontinuity. The accuracy of the method is compared to commercial software based on the finite element method. The proposed method is exclusively competent for the frequency analysis of rounded or arbitrarily shaped metallic discontinuities.
Novel saturated flow boiling heat transfer correlation for R32 refrigerant
(AIAA International, 2022) Turgut Oğuz Emrah; Asker Mustafa; Genceli Hadi; Çoban Mustafa Turhan
[No abstract available]
Multi-objective particle swarm optimization of the k-type shell and tube heat exchanger (case study)
(Yildiz Technical University, 2021) Nadi M.; Ehyaei M.A.; Ahmadi A.; Turgut, Oğuz Emrah
This paper investigated optimization of two objectives function include the total amount of heat transfer between two mediums and the total cost of shell and tube heat exchanger. The study was carried out for k-type heat exchanger of the cryogenic unit of gas condensates by multiple objective particle swarm optimization. Six decision variables including pipe pitch ratio, pipe diameter, pipe number, pipe length, baffle cut ratio, and baffle distance ratio were taking into account to conduct this simulation-based research. The results of mathematical modeling confirmed the actual results (data collected from the evaporator unit of the Tehran refinery’s absorption chiller). The optimization results revealed that the two objective functions of heat transfer rate and the total cost were in contradiction with each other. The results of the sensitivity analysis showed that with change in the pitch ratio from 1.25 to 2, the amount of heat transfer was reduced from 420 to 390 kW about 7.8%. Moreover, these variations caused reduction in cost function from 24,500 to 23,500 $, less than 1%. On the other hand, an increase in pipe length from 3 to 12 meters, the heat transfer rate raised from 365 to 415 kW by 13.7%, while the cost increased from 20,000$ to 24500$ about 22%. © 2021. All rights reserved.
Parameter estimation of the wind speed distribution model by dragonfly algorithm
(Gazi Universitesi, 2023) Köse, Bayram; Aygün, Hilmi; Pak, Semih
In order to meet the increasing energy demand and to solve environmental problems, the interest in renewable energy sources continues with technology development studies and economic investments. Various methods are used to determine and estimate sustainable and renewable energy sources. Probability distribution functions are used in wind characterization and potential calculation of wind energy. In this study, the Dragonfly Algorithm (DA) is proposed to estimate the Weibull probability distribution function (Wpdf) parameters used in wind speed modeling and the two-component mixed Weibull distribution parameters used in modeling non-single peak wind speed data. The performance of the proposed method has been evaluated by comparing not only the classical methods which are the moment method (MM) and the least squares method (LSM) but also metaheuristic optimization algorithms which are Genetic Algorithm (GA) and Particle Swarm Optimization (PSO). Determination coefficient (R2) and root mean square error (RMSE) were used to evaluate the performance of these parameter estimation methods. Data obtained from 6 measurement stations were used in the study. According to the performance criteria, the two-component Weibull distribution was found to be more effective at all stations compared to the Weibull distribution model. It has been concluded that the proposed DA algorithm can be used effectively for parameter estimation in wind speed modeling. © 2023 Gazi Universitesi Muhendislik-Mimarlik. All rights reserved.
Local search enhanced Aquila optimization algorithm ameliorated with an ensemble of Wavelet mutation strategies for complex optimization problems
(Elsevier, 2023) Turgut, Oğuz Emrah; Turgut, Mert Sinan
Aquila Optimization Algorithm (AQUILA) is a newly emerged metaheuristic optimizer for solving global optimization problems, which is based on intrinsic hunting behaviors of the foraging aquila individuals. However, this stochastic optimization method suffers from some algorithm-specific drawbacks, such as premature convergence to the local optimum points over the search hyperspace due to the lack of solution diversity in the population. To conquer this algorithmic deficiency, an ensemble of Wavelet mutation operators has been implemented into the standard AQUILA to enhance the explorative capabilities of the algorithm by diversifying the search domain as much as possible. Furthermore, a brand-new local search scheme empowered by the synergetic interactions of elite opposition-based learning and a simple-yet-effective exploitative manipulation equation is introduced into the base AQUILA to intensify on the previously visited promising regions. The proposed learning schemes are stochastically applied to the obtained solutions from the base Aquila algorithm to refine the overall solution quality and amend the premature convergence problem. It is also aimed to investigate whether the collective application of Wavelet mutation operators with different types entails a significant improvement in the general search effectivity of the algorithm rather than their individual efforts. Numerical experiments made on a suite of unconstrained unimodal and multimodal benchmark functions reveal that this hybridization with AQUILA has improved the general solution accuracy and stability to very high standards, outperforming its contemporary counterparts in the comparative statistical analysis. Furthermore, an exhaustive benchmark analysis has been performed on fourteen constrained real-world complex engineering problems.(c) 2022 International Association for Mathematics and Computers in Simulation (IMACS). Published by Elsevier B.V. All rights reserved.
Chaotic quasi-oppositional arithmetic optimization algorithm for thermo-economic design of a shell and tube condenser running with different refrigerant mixture pairs
(Springer London Ltd, 2022) Turgut, Mert Sinan; Turgut, Oğuz Emrah; Abualigah, Laith
This theoretical research study proposes a novel Chaotic Quasi-Oppositional Arithmetic Optimization Algorithm (COAOA) for thermo-economic optimization of a shell and tube condenser working with refrigerant mixtures. Arithmetic Optimization Algorithm (AOA) is a recently emerged metaheuristic algorithm considering different mathematical operators to optimize the candidate solutions over a wide range of search domains. The effectiveness the COAOA is assessed by applying it to a set of benchmark optimization problems and comparing the obtained solutions with that of the original AOA and its quasi-oppositional variant. The COAOA has been employed to acquire the minimum value of the total annual cost of the shell and tube condenser by iteratively varying nine decision variables of mass flow rate, shell diameter, the tube inside diameter, tube length, number of tube passes, tube layout, tube pitch ratio, the total number of baffles, and diameter ratio. Three different case studies are solved using different refrigerant pairs used for in-tube flow to show the proposed metaheuristic optimizer's efficiency and effectivity on real-world mixed-integer optimization problem. Optimal results retrieved for different mixture pairs with varying mass fractions are compared with each other, and parametric configuration yielding the minimum total cost is decided. Finally, a comprehensive sensitivity analysis is performed to investigate the influences of the design variables over the considered problem objective. Overall analysis results indicate that COAOA can be an excellent optimizer to obtain a shell and tube condenser's optimal configuration within a reasonable computation time.
Dielectric measurement via partially filled and short-circuited circular waveguides
(Ieee, 2020) Aydoğan, Ahmet; Sağlam, Serkan; Akleman, Funda
In this study, the dielectric measurement of homogeneous materials via short-circuited circular waveguides is examined. The proposed method is numerically tested against homogeneous materials with varying permittivities, as well as in different shapes and positions. Synthetic data is generated via the moment method solution of the related integral equation. In the inversion algorithm, Newton-Raphson method with multi-frequency data is used. The complex part of the material is modeled in a linear fashion.
COVID-19 detection with severity level analysis using the deep features, and wrapper-based selection of ranked features
(Wiley, 2021) Öksüz, Coşku; Urhan, Oğuzhan; Güllü, Mehmet Kemal
The SARS-COV-2 virus, which causes COVID-19 disease, continues to threaten the whole world with its mutations. Many methods developed for COVID-19 detection are validated on the data sets generally including severe forms of the disease. Since the severe forms of the disease have prominent signatures on X-ray images, the performance to be achieved is high. To slow the spread of the disease, effective computer-assisted screening tools with the ability to detect the mild and the moderate forms of the disease that do not have prominent signatures are needed. In this work, various pretrained networks, namely GoogLeNet, ResNet18, SqueezeNet, ShuffleNet, EfficientNetB0, and Xception, are used as feature extractors for the COVID-19 detection with severity level analysis. The best feature extraction layer for each pre-trained network is determined to optimize the performance. After that, features obtained by the best layer are selected by following a wrapper-based feature selection strategy using the features ranked based on Laplacian scores. The experimental results achieved on two publicly available data sets including all the forms of COVID-19 disease reveal that the method generalized well on unseen data. Moreover, 66.67%, 90.32%, and 100% sensitivity are obtained in the detection of mild, moderate, and severe cases, respectively.
Brain tumor classification using the fused features extracted from expanded tumor region
(Elsevier Sci Ltd, 2022) Öksüz, Coşku; Urhan, Oğuzhan; Güllü, Mehmet Kemal
In this study, a brain tumor classification method using the fusion of deep and shallow features is proposed to distinguish between meningioma, glioma, pituitary tumor types and to predict the 1p/19q co-deletion status of LGG tumors. Brain tumors can be located in a different region of the brain, and the texture of the surrounding tissues may also vary. Therefore, the inclusion of surrounding tissues into the tumor region (ROI expansion) can make the features more distinctive. In this work, pre-trained AlexNet, ResNet-18, GoogLeNet, and ShuffleNet networks are used to extract deep features from the tumor regions including its surrounding tissues. Even though the deep features are extremely important in classification, some low-level information regarding tumors may be lost as the network deepens. Accordingly, a shallow network is designed to learn low-level information. Next, in order to compensate the information loss, deep features and shallow features are fused. SVM and k-NN classifiers are trained using the fused feature sets. Experimental results achieved on two publicly available data sets demonstrate that using the feature fusion and the ROI expansion at the same time improves the average sensitivity by about 11.72% (ROI expansion: 8.97%, feature fusion: 2.75%). These results confirm the assumption that the tissues surrounding the tumor region carry distinctive information. Not only that, the missing low-level information can be compensated thanks to the feature fusion. Moreover, competitive results are achieved against state-of-the-art studies when the ResNet-18 is used as the deep feature extractor of our classification framework.
Ensemble-LungMaskNet: Automated lung segmentation using ensembled deep encoders
(Institute of Electrical and Electronics Engineers Inc., 2021) Öksüz, Coşku; Urhan, Oğuzhan; Güllü, Mehmet Kemal
Automated lung segmentation has importance because it gives clues about several diseases to the experts. It is the step that comes before further detailed analyses of the lungs. However, segmentation of the lungs is a challenging task since the opacities and consolidations are caused by various lung diseases. As a result, the clarity of the borders of the lungs may be lost which makes the segmentation task difficult. The presence of various medical equipment such as cables in the image is another factor that makes segmentation difficult. Therefore, it is a necessity to develop methods that can handle such situations. Learning the most useful patterns related to various diseases is possible with deep learning methods. Unlike conventional methods, learning the patterns improves the generalization ability of the models on unseen data. For this purpose, a deep segmentation framework including ensembles of pre-trained lightweight networks is proposed for lung region segmentation in this work. The experimental results achieved on two publicly available data sets demonstrate the effectiveness of the proposed framework. © 2021 IEEE.

Güncel Gönderiler