Electrical - Electronic Engineering / Elektrik - Elektronik Mühendisliği
Permanent URI for this collectionhttps://hdl.handle.net/11147/11
Browse
49 results
Search Results
Article Citation - WoS: 5Citation - Scopus: 7Adaptive Resizer-Based Transfer Learning Framework for the Diagnosis of Breast Cancer Using Histopathology Images(Springer, 2023) Düzyel, Okan; Çatal, Mehmet Sergen; Kayan, Ceyhun Efe; Sevinç, Arda; Gümüş, AbdurrahmanBreast cancer is a major global health concern, and early and accurate diagnosis is crucial for effective treatment. Recent advancements in computer-assisted prediction models have facilitated diagnosis and prognosis using high-resolution histopathology images, which provide detailed information on cancerous tissue. However, these high-resolution images often require resizing, leading to potential data loss. In this study, we demonstrate the effect of a learnable adaptive resizer for breast cancer classification using the BreakHis dataset. Our approach incorporates the adaptive resizer with various convolutional neural network models, including VGG16, VGG19, MobileNetV2, InceptionResnetV2, DenseNet121, DenseNet201, and EfficientNetB0. Despite producing visually less appealing images, the learnable resizer effectively improves classification performance. DenseNet201, when jointly trained with the adaptive resizer, achieves the highest accuracy of 98.96% for input images of 448x448 resolution. Our experimental results demonstrate that the adaptive resizer performs better at a magnification factor of 40x compared to higher magnifications. While its effectiveness becomes less pronounced as image resolution increases to 100x, 200x, and 400x, the adaptive resizer still outperforms bilinear interpolation. In conclusion, this study highlights the potential of adaptive resizers in enhancing performance for medical image classification. By outperforming traditional image resizing methods, our work contributes to the advancement of deep neural networks in the field of breast cancer diagnostics.Article Citation - WoS: 2Citation - Scopus: 2Subwavelength Thickness Characterization of Curved Dielectric Films Exploiting Spatially Structured Entangled Photons(Optica Publishing Group, 2023) Ataç, Enes; Dinleyici, Mehmet SalihPrecise determination of thin dielectric film optical properties is a critical issue for fiber optic sensor technologies. However, conventional methods for the optical characterization of these films not only are generally complex and tedious processes on curved surfaces but also require well-calibrated and overly sophisticated devices. We, on the other hand, propose a novel and practical quantum-based phase diffraction scheme to characterize the thickness of ultra-thin transparent dielectric films coated on an optical fiber beyond the classical diffraction limits in this paper. The approach is implemented by evaluating the effect of thickness variations on the highly visible two-photon diffraction pattern's zero crossings and amplitudes. The mathematical model and numerical simulations con-tribute to a better understanding of how the spatially structured entangled photons improve thickness precision with the help of intensity correlations and a confocal aperture. To prove the impact of the proposed system, it is compared with the classical phase diffraction method in the literature via simulations. According to the results, the thickness of the transparent dielectric films can be accurately estimated below one-twentieth of the wavelength of interest. & COPY; 2023 Optica Publishing GroupArticle Citation - WoS: 3Citation - Scopus: 5Event Distortion-Based Clustering Algorithm for Energy Harvesting Wireless Sensor Networks(Springer, 2021) Al-Qamaji, Ali; Atakan, BarışWireless sensor networks (WSNs) consist of compact deployed sensor nodes which collectively report their sensed readings about an event to the Base Station (BS). In WSNs, due to the dense deployment, sensor readings can be spatially correlated and it is nonessential to transmit all their readings to the BS. Therefore, for more energy efficient, it is vital to choose which sensor node should report their sensed readings to the BS. In this paper, the event distortion-based clustering (EDC) algorithm is proposed for the spatially correlated sensor nodes. Here, the sensor nodes are assumed to harvest energy from ambient electromagnetic radiation source. The EDC algorithm allows the energy-harvesting sensor nodes to select and eliminate nonessential nodes while maintain an acceptable level of distortion at the BS. To measure the reliability, a theoretical framework of the distortion function is first derived for both single-hop and two-hop communication scenarios. Then, based on the derived theoretical framework, the EDC algorithm is introduced. Through extensive simulations, the performance of the EDC algorithm is evaluated in terms of achievable distortion level, number of alive nodes and harvested energy levels. As a result, EDC algorithm can successfully exploit both the spatial correlation and energy harvesting to improve the energy efficiency while preserving an acceptable level of distortion. Furthermore, the performance comparisons reveal that the two-hop communication model outperforms the single-hop model in terms of the distortion and energy-efficiency.Article Citation - WoS: 4Citation - Scopus: 5Toward Safe and High-Performance Human-Robot Collaboration Via Implementation of Redundancy and Understanding the Effects of Admittance Term Parameters(Cambridge University Press, 2022) Kanık, Mert; Ayit, Orhan; Dede, Mehmet İsmet Can; Tatlıcıoğlu, EnverSummary Today, demandsin industrial manufacturing mandate humans to work with large-scale industrial robots, and this collaboration may result in dangerous conditions for humans. To deal with this situation, this work proposes a novel approach for redundant large-scale industrial robots. In the proposed approach, an admittance controller is designed to regulate the interaction between the end effector of the robot and the human. Additionally, an obstacle avoidance algorithm is implemented in the null space of the robot to prevent any possible unexpected collision between the human and the links of the robot. After safety performance of this approach is verified via simulations and experimental studies, the effect of the parameters of the admittance controller on the performance of collaboration in terms of both accuracy and total human effort is investigated. This investigation is carried out via 8 experiments by the participation of 10 test subjects in which the effect of different admittance controller parameters such as mass and damper are compared. As a result of this investigation, tuning insights for such parameters are revealed.Article Citation - WoS: 2Citation - Scopus: 2Deep Learning Based Adaptive Bit Allocation for Heterogeneous Interference Channels(Elsevier, 2021) Aycan, Esra; Özbek, Berna; Le Ruyet, DidierThis paper proposes an adaptive bit allocation scheme by using a fully connected (FC) deep neural network (DNN) considering imperfect channel state information (CSI) for heterogeneous networks. Achieving an accurate CSI has a crucial role on the system performance of the heterogeneous networks. Different quantization techniques have been employed to reduce the feedback overhead. However, the system performance cannot increase linearly with the number of bits increasing exponentially. Since optimizing the total number of bits is too complex for the entire network, an initial step is performed to distribute the bits to each cell in the conventional method. Then, the distributed bits are further allocated to each channel optimally. In order to enable direct allocation for the entire network, a FC-DNN based method is presented in this study. The optimized number of bits can be directly obtained for a different number of bits and scenarios by the proposed approach. The simulations are performed by using various scenarios with different allocation schemes. The performance results show that the DNN based method achieves a closer performance to the conventional approach. (C) 2021 Elsevier B.V. All rights reserved.Article Time-Efficient Evaluation of Adaptation Algorithms for Dash With Svc: Dataset, Throughput Generation and Stream Simulator(Springer, 2021) Çalı, Mehmet; Özbek, NükhetBitrate adaptation algorithms have received considerable attention recently. In order to evaluate these algorithms objectively, multiple DASH datasets have been proposed. However, only few of them are compatible to SVC-based adaptation algorithms. Apart from the dataset, to fully implement and evaluate an adaptation algorithm, many time-consuming steps are required such as MPD parser design, adaptation logic design and network environment setup. In this paper, a dash simulator which assesses the performance of SVC-based adaptation algorithms without the requirement of any additional implementation steps is proposed. Also, an SVC dataset that includes both CBR and VBR encoded videos is designed. Demonstration is performed as evaluation of an SVC-based adaptation algorithm under several throughput scenarios using the designed dataset. Results show that the proposed system considerably reduces time requirement compared to real-time assessment. Dataset, throughput generation tool and simulator are all publicly available so that the researchers can test their implementation and compare with the results presented in this paper.Article Citation - WoS: 4Citation - Scopus: 5Hybrid Beamforming Strategies for Secure Multicell Multiuser Mmwave Mimo Communications(Elsevier, 2021) Özbek, Berna; Erdoğan, Oğulcan; Busari, Sherif A.; Gonzalez, JonathanOver the last decade, many advancements have been made in the field of wireless communications. Among the major technology enablers being explored for the beyond fifth-generation (B5G) networks at the physical layer (PHY), a great deal of attention has been focused on millimeter-wave (mmWave) communications, massive multiple-input multiple-output (MIMO) antenna systems and beamforming techniques. These enablers bring to the forefront great opportunities for enhancing the performance of B5G networks, concerning spectral efficiency, energy efficiency, latency, and reliability. The wireless communication is prone to information leakage to the unintended nodes due to its open nature. Hence, the secure communication is becoming more critical in the wireless networks. To address this challenge, the concept of Physical Layer Security (PLS) is explored in the literature. In this paper, we examine the mmWave transmission through linear beamforming techniques for PLS based systems. We propose the secure multiuser (MU) MIMO mmWave communications by employing hybrid beamforming at the base stations (BSs), legitimate users and eavesdroppers. Using three Dimensional (3D) mmWave channel model for each node, we utilize the artificial noise (AN) beamforming to jam the transmission of eavesdropper and to enhance the secrecy rate. The secrecy performance on multicell mmWave MU-MIMO downlink communications is demonstrated to reveal the key points directly related to the system security for B5G wireless systems. (C) 2021 Elsevier B.V. All rights reserved.Article Citation - WoS: 5Citation - Scopus: 7Fast Texture Classification of Denoised Sar Image Patches Using Glcm on Spark(Türkiye Klinikleri Journal of Medical Sciences, 2020) Özcan, Caner; Ersoy, Okan; Oğul, İskender ÜlgenClassification of a synthetic aperture radar (SAR) image is an essential process for SAR image analysis and interpretation. Recent advances in imaging technologies have allowed data sizes to grow, and a large number of applications in many areas have been generated. However, analysis of high-resolution SAR images, such as classification, is a time-consuming process and high-speed algorithms are needed. In this study, classification of high-speed denoised SAR image patches by using Apache Spark clustering framework is presented. Spark is preferred due to its powerful open-source cluster-computing framework with fast, easy-to-use, and in-memory analytics. Classification of SAR images is realized on patch level by using the supervised learning algorithms embedded in the Spark machine learning library. The feature vectors used as the classifier input are obtained using gray-level cooccurrence matrix which is chosen to quantitatively evaluate textural parameters and representations. SAR image patches used to construct the feature vectors are first applied to the noise reduction algorithm to obtain a more accurate classification accuracy. Experimental studies were carried out using naive Bayes, decision tree, and random forest algorithms to provide comparative results, and significant accuracies were achieved. The results were also compared with a state-of-the-art deep learning method. TerraSAR-X images of high-resolution real-world SAR images were used as data.Article Citation - WoS: 4Ramcess 2.x Framework-Expressive Voice Analysis for Realtime and Accurate Synthesis of Singing(Springer Verlag, 2008) d'Alessandro, Nicolas; Babacan, Onur; Bozkurt, Barış; Dubuisson, Thomas; Holzapfel, Andre; Kessous, Loic; Vlieghe, MaximeIn this paper we present the work that has been achieved in the context of the second version of the RAMCESS singing synthesis framework. The main improvement of this study is the integration of new algorithms for expressive voice analysis, especially the separation of the glottal source and the vocal tract. Realtime synthesis modules have also been refined. These elements have been integrated in an existing digital instrument: the HANDSKETCH 1.X, a bimanual controller. Moreover this digital instrument is compared to existing systems.Article Traffic Aware Cell Selection Algorithm for Tetra Trunk Based Professional Mobile Radio(Springer Verlag, 2019) Özbek, Berna; Karataş, Azad; Bardak, Erinç Deniz; Sönmez, İlkerLoad balancing and traffic management are the critical needs in cell selection decision for a better and seamless communication demands in professional mobile radios. For the cases where cell selection algorithms do not consider the traffic load, there may be call drops due to the congestion in networks or longer call setup times for the users. These undesired consequences can cause dramatic quality degradation especially for the emergency cases or public safety services. In this paper, we propose two algorithms for Tetra Trunk based professional mobile radios by considering both traffic load and received signal strength indication (RSSI) in order to reduce the significant delays while establishing transmissions. The proposed full set cell selection algorithm is prioritized to reduce the call setup time for the mobile users and the proposed reduced set cell selection algorithm is focused on minimizing the number of RSSI measurements which causes significant delay in practical professional mobile radio. We illustrate the performance results in terms of delay for Tetra Trunk based professional mobile radio.