36 sonuçlar
Arama Sonuçları
Listeleniyor 1 - 10 / 36
Yayın Developing an efficient deep neural network for automatic detection of COVID-19 using chest X-ray images(Elsevier B.V., 2021-06) Sheykhivand, Sobhan; Mousavi, Zohreh; Mojtahedi, Sina; Yousefi Rezaii, Tohid; Farzamnia, Ali; Meshgini, Saeed; Saad, IsmailThe novel coronavirus (COVID-19) could be described as the greatest human challenge of the 21st century. The development and transmission of the disease have increased mortality in all countries. Therefore, a rapid diagnosis of COVID-19 is necessary to treat and control the disease. In this paper, a new method for the automatic identification of pneumonia (including COVID-19) is presented using a proposed deep neural network. In the proposed method, the chest X-ray images are used to separate 2–4 classes in 7 different and functional scenarios according to healthy, viral, bacterial, and COVID-19 classes. In the proposed architecture, Generative Adversarial Networks (GANs) are used together with a fusion of the deep transfer learning and LSTM networks, without involving feature extraction/selection for classification of pneumonia. We have achieved more than 90% accuracy for all scenarios except one and also achieved 99% accuracy for separating COVID-19 from healthy group. We also compared our deep proposed network with other deep transfer learning networks (including Inception-ResNet V2, Inception V4, VGG16 and MobileNet) that have been recently widely used in pneumonia detection studies. The results based on the proposed network were very promising in terms of accuracy, precision, sensitivity, and specificity compared to the other deep transfer learning approaches. Depending on the high performance of the proposed method, it can be used during the treatment of patients.Yayın Multi-task learning on mental disorder detection, sentiment detection and emotion detection(Işık Üniversitesi, 2024-02-12) Armah, Courage; Dehkharghani, Rahim; Işık Üniversitesi, Lisansüstü Eğitim Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans Programı; Işık University, School of Graduate Studies, Computer Science Engineering Master ProgramSuicidal behavior is a global cause of life-threatening injury and most of the time, death. Mental disorders such as depression, anxiety, and bipolar are prevalent among the youth in recent decades. Social media are popular platforms for individuals to post their thoughts and feelings on. Extracting people’s sentiments and feelings from such online platforms would help detect mental disorders of the users to treat them before it becomes too late. This thesis investigates the use of multi-task learning systems and single-task learning techniques to estimate behaviors and mental states for early diagnosis. I used data mined from Reddit, one of the popular social media platforms that provides anonymity. Anonymity increases the chances of individuals sharing what they truly feel in their real life. The obtained results by the proposed approaches open new doors to the understanding of how multi-task systems can increase the performance of text classification problems such as depression detection, emotion detection, and sentiment analysis, trained together in a multi-task learning network when compared to their training in isolation in a single-task learning network. We used the SWMH dataset, already labeled by 5 different depression labels (depression, anxiety, suicide, bipolar, and off my chest) and then added emotion and polarity labels to it and made it publicly available for researchers in the literature. The obtained results in this study are also comparable to other approaches in the field.Yayın Deep learning techniques for building density estimation from remotely sensed imagery(Işık Üniversitesi, 2019-04-05) Süberk, Nilay Tuğçe; Ateş, Hasan Fehmi; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Elektronik Mühendisliği Yüksek Lisans ProgramıThis thesis is about point-wise estimation of building density on the remote sensing optical imageries by applying deep learning methods. The goal of the project is to reduce mean square error of the estimated density by applying architectural modi?cations on the deep learning network and using augmented training data. Recently, deep learning is one of popular ?eld of science and convolutional neural networks (CNNs) are well-known deep neural network. Recent studies indicate that some of the convolutional neural networks are highly e?ective in large scale image works such as recognition, semantic segmentation. There has been limited research in using deep networks to learn urbanization characteristics from remote sensing images. Remote sensing images could be used for regression problems and building density estimation is one of them. Building density information provides knowledge for real estate agents and urban planners, estimating disaster risk areas, environment protection and resource allocation. Our method provides a cheap and fast solution to these needs when there is no cadastral information. The main objective of this thesis is to achieve fast and accurate local building density estimation using high resolution remote sensing images. Deep learning methods based on CNN are applied in this project. Pre-trained visual geometry group (VGG-16) and fully convolutional network (FCN) are tested as convolutional neural network. We tested three di?erent modi?ed networks and then applied data augmentation in the train data to reduce mean square error value. The networks that we have performed simpli?ed original VGG-16 network for regression, VGG-16 network with sigmoid layer added and simpli?ed VGG-16 network with sigmoid layer. The best result (lowest mean square error) is obtained from sigmoid layer added VGG-16 network with data augmentation. Sigmoid layer added VGG-16 network gives us (?0,084) RMSE on building density estimation with the augmented train dataset. Original VGG-16 network gives (?0,105) RMSE, sigmoid layer added VGG-16 network gives (?0,095) RMSE and sigmoid layer added simpli?ed VGG-16 network gives (?0,090) RMSE on building density estimation with the small train dataset. FCN is one of the ideal network for classi?cation tasks so we have also applied fully convolutional network result to compare our results with its result. We have modi?ed the network to perform building density estimation in addition to semantic segmentation. The root mean square error of FCN is (?0,084) and our best result (lowest mean square error) is also (?0,084) RMSE at the same iteration number. Our results show that fast and accurate building density estimation is possible by using vanilla CNNs. Sigmoid layer addition, simpli?cation of the network for small dataset and data augmentation improves accuracy in the regression. Data augmentation is the most e?ective method to reduce RMSE in this thesis.Yayın Convolutional attention network for MRI-based Alzheimer's disease classification and its interpretability analysis(IEEE, 2021-09-17) Türkan, Yasemin; Tek, Faik BorayNeuroimaging techniques, such as Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET), help to identify Alzheimer's disease (AD). These techniques generate large-scale, high-dimensional, multimodal neuroimaging data, which is time-consuming and difficult to interpret and classify. Therefore, interest in deep learning approaches for the classification of 3D structural MRI brain scans has grown rapidly. In this research study, we improved the 3D VGG model proposed by Korolev et al. [2]. We increased the filters in the 3D convolutional layers and then added an attention mechanism for better classification. We compared the performance of the proposed approaches for the classification of Alzheimer's disease versus mild cognitive impairments and normal cohorts on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. We observed that both the accuracy and area under curve results improved with the proposed models. However, deep neural networks are black boxes that produce predictions that require further explanation for medical usage. We compared the 3D-data interpretation capabilities of the proposed models using four different interpretability methods: Occlusion, 3D Ultrametric Contour Map, 3D Gradient-Weighted Class Activation Mapping, and SHapley Additive explanations (SHAP). We observed that explanation results differed in different network models and data classes.Yayın CNN-Based deep learning architecture for electromagnetic imaging of rough surface profiles(IEEE, 2022-10) Aydın, İzde; Budak, Güven; Sefer, Ahmet; Yapar, AliA convolutional neural network (CNN) based deep learning (DL) technique for electromagnetic imaging of rough surfaces separating two dielectric media is presented. The direct scattering problem is formulated through the conventional integral equations and the synthetic scattered field data is produced by a fast numerical solution technique which is based on Method of Moments (MoM). Two different special CNN architectures are designed and implemented for the solution of the inverse rough surface imaging problem wherein both random and deterministic rough surface profiles can be imaged. It is shown by a comprehensive numerical analysis that the proposed deep-learning (DL) inversion scheme is very effective and robust.Yayın Convolutional neural network (CNN) algorithm based facial emotion recognition (FER) system for FER-2013 dataset(IEEE, 2022-11-18) Ezerceli, Özay; Eskil, Mustafa TanerFacial expression recognition (FER) is the key to understanding human emotions and feelings. It is an active area of research since human thoughts can be collected, processed, and used in customer satisfaction, politics, and medical domains. Automated FER systems had been developed and have been used to recognize humans’ emotions but it has been a quite challenging problem in machine learning due to the high intra-class variation. The first models were using known methods such as Support Vector Machines (SVM), Bayes classifier, Fuzzy Techniques, Feature Selection, Artificial Neural Networks (ANN) in their models but still, some limitations affect the accuracy critically such as subjectivity, occlusion, pose, low resolution, scale, illumination variation, etc. The ability of CNN boosts FER accuracy. Deep learning algorithms have emerged as the greatest way to produce the best results in FER in recent years. Various datasets were used to train, test, and validate the models. FER2013, CK+, JAFFE and FERG are some of the most popular datasets. To improve the accuracy of FER models, one dataset or a mix of datasets has been employed. Every dataset includes limitations and issues that have an impact on the model that is trained for it. As a solution to this problem, our state-of-the-art model based on deep learning architectures, particularly convolutional neural network architectures (CNN) with supportive techniques has been implemented. The proposed model achieved 93.7% accuracy with the combination of FER2013 and CK+ datasets for FER2013.Yayın A theoretical comparison of ResNet and DenseNet architectures on the subject of shoreline extraction(Işık Üniversitesi, 2020-09-23) Ecevit, Mert İlhan; Çavdaroğlu, Gülsüm Çiğdem; Işık Üniversitesi, Fen Bilimleri Enstitüsü, Enformasyon Teknolojileri Yüksek Lisans ProgramıToday's Deep Learning technologies provides numerous approaches on the subject of convolutional networks. These approaches serve researchers to train datasets and generate wanted results from these datasets. Each CNN architecture has its own strong points and weak sides. Because of this situation a comparison between these architectures is a valuable asset. Image processing is a method that is frequently used to process remotely sensed data in remote sensing studies.. Between current architectures, RESNET and DENSENET architectures are chosen to be used by Dr. Çavdaroğlu for her project on TÜBİTAK. The result of this comparison will be used in that project in order to apply most ecient architecture. This thesis is written to draw outlines of RESNET and DENSENET and create a foresight for further projects which can be supported by this thesis. In order to achieve an accurate image recognition process in remote sensing domain, a preliminary research is requisite. As a research thesis this work serves the purpose of learning manner of works, performance indicators of RESNET and DENSENET convolutional networks. The result of this research will create a baseline for an academical project. At the other hand, comparison of these two convolutional network approaches provides information to decide which approach is more suitable for remote sensing projects depending upon the subject of the project. For future works on Remote Sensing this thesis work will serve a guideline and reason for preference. The presented thesis work has been developed as the technical feasibility of the 3501 TÜBITAK Project named "Uydu Görüntülerinden Kıyı Sınırlarının Derin Öğrenme Yöntemleri ile Otomatik Çıkarımı", applied by Dr. G. Çiğdem Çavdaroğlu, and the thesis results will be applied within the scope of the Project after the project acceptance.Yayın Adaptive convolution kernel for artificial neural networks(Academic Press Inc., 2021-02) Tek, Faik Boray; Çam, İlker; Karlı, DenizMany deep neural networks are built by using stacked convolutional layers of fixed and single size (often 3 × 3) kernels. This paper describes a method for learning the size of convolutional kernels to provide varying size kernels in a single layer. The method utilizes a differentiable, and therefore backpropagation-trainable Gaussian envelope which can grow or shrink in a base grid. Our experiments compared the proposed adaptive layers to ordinary convolution layers in a simple two-layer network, a deeper residual network, and a U-Net architecture. The results in the popular image classification datasets such as MNIST, MNIST-CLUTTERED, CIFAR-10, Fashion, and ‘‘Faces in the Wild’’ showed that the adaptive kernels can provide statistically significant improvements on ordinary convolution kernels. A segmentation experiment in the Oxford-Pets dataset demonstrated that replacing ordinary convolution layers in a U-shaped network with 7 × 7 adaptive layers can improve its learning performance and ability to generalize.Yayın Automatic propbank generation for Turkish(Incoma Ltd, 2019-09) Ak, Koray; Yıldız, Olcay TanerSemantic role labeling (SRL) is an important task for understanding natural languages, where the objective is to analyse propositions expressed by the verb and to identify each word that bears a semantic role. It provides an extensive dataset to enhance NLP applications such as information retrieval, machine translation, information extraction, and question answering. However, creating SRL models are difficult. Even in some languages, it is infeasible to create SRL models that have predicate-argument structure due to lack of linguistic resources. In this paper, we present our method to create an automatic Turkish PropBank by exploiting parallel data from the translated sentences of English PropBank. Experiments show that our method gives promising results. © 2019 Association for Computational Linguistics (ACL).Yayın Malaria parasite detection with deep transfer learning(IEEE, 2018-12-06) Var, Esra; Tek, Faik BorayThis study aims to automatically detect malaria parasites (Plasmodium sp) on images taken from Giemsa stained blood smears. Deep learning methods provide limited performance when sample size is low. In transfer learning, visual features are learned from large general data sets, and problem-specific classification problem can be solved successfully in restricted problem specific data sets. In this study, we apply transfer learning method to detect and classify malaria parasites. We use a popular pre-trained CNN model VGG19. We trained the model for 20 epoch on 1428 P Vivax, 1425 P Ovule, 1446 E Falciparum, 1450 P Malariae and 1440 non-parasite samples. The transfer learning model achieves %80, %83, %86, %75 precision and 83%, 86%, 86%, 79% f-measure on 19 test images.
