10 sonuçlar
Arama Sonuçları
Listeleniyor 1 - 10 / 10
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 Derin öznitelikler ile anlambilimsel görüntü bölütleme(Institute of Electrical and Electronics Engineers Inc., 2018-07-05) Sünetci, Sercan; Ateş, Hasan FehmiDerin evrişimsel sinir ağları (ESA) pek çok sınıflandırma probleminde olduğu gibi anlambilimsel görüntü bölütlemede de çok ciddi başarı göstermiştir. Fakat derin ağların eğitilmesi hem zaman alıcıdır hem de geniş bir eğitim veri kümesine ihtiyaç duymaktadır. Bir veri kümesinde eğitilen ağın başka bir görev ya da veri kümesine uygulanabilmesi için transfer öğrenme ile yeniden eğitilmesi gerekmektedir. Transfer öğrenmeye alternatif olarak ağ katmanlarından çıkarılan öznitelik vektörleri doğrudan sınıflandırma amaçlı kullanılabilir. Bu bildiride genel ESA mimarilerinden elde edilen özniteliklerin eğitim gerektirmeyen bir görüntü etiketleme yönteminde kullanılmasının sınıflandırma başarımına katkısı incelenmiştir. Derin ağlarda ‘öğrenilmiş’ öznitelikler ile SIFT gibi ‘el yapımı’ özniteliklerin birlikte kullanılmasının etiketleme doğruluğunu artırdığı gösterilmiştir. Varolan ön eğitimli ağların kullanılması sayesinde önerilen yaklaşım herhangi bir veri kümesinde yeniden eğitime gerek olmadan kolayca uygulanabilmektedir. Önerilen yöntem iki veri kümesinde test edilmiş ve etiketleme doğruluğu benzer yöntemlerle karşılaştırmalı olarak sunulmuştur.Yayın Uyarlanır yerel bağlı katman kullanan dikkat tabanlı derin ağ ile sesli komut tanıma(Institute of Electrical and Electronics Engineers Inc., 2020-10-05) Turkan, Yasemin; Tek, Faik BoraySesli komut tanıma insan-makine ara yüzüyle ilişkili aktif bir araştırma konusudur. Dikkat tabanlı derin ağlar ile bu tür problemler başarılı bir şekilde çözülebilmektedir. Bu çalışmada, var olan bir dikkat tabanlı derin ağ yöntemi, uyarlanır yerel bağlı (odaklanan) katman kullanılarak daha da geliştirilmiştir. Orijinal yönteminde sınandığı Google ve Kaggle sesli komut veri setlerinde karşılaştırmalı olarak yapılan deneylerde önerdiğimiz uyarlanır yerel bağlı katman kullanan dikkat tabanlı ağın tanıma doğruluğunu %2.6 oranında iyileştirdiği gözlemledik.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 Unreasonable effectiveness of last hidden layer activations for adversarial robustness(Institute of Electrical and Electronics Engineers Inc., 2022) Tuna, Ömer Faruk; Çatak, Ferhat Özgür; Eskil, Mustafa TanerIn standard Deep Neural Network (DNN) based classifiers, the general convention is to omit the activation function in the last (output) layer and directly apply the softmax function on the logits to get the probability scores of each class. In this type of architectures, the loss value of the classifier against any output class is directly proportional to the difference between the final probability score and the label value of the associated class. Standard White-box adversarial evasion attacks, whether targeted or untargeted, mainly try to exploit the gradient of the model loss function to craft adversarial samples and fool the model. In this study, we show both mathematically and experimentally that using some widely known activation functions in the output layer of the model with high temperature values has the effect of zeroing out the gradients for both targeted and untargeted attack cases, preventing attackers from exploiting the model's loss function to craft adversarial samples. We've experimentally verified the efficacy of our approach on MNIST (Digit), CIFAR10 datasets. Detailed experiments confirmed that our approach substantially improves robustness against gradient-based targeted and untargeted attack threats. And, we showed that the increased non-linearity at the output layer has some ad-ditional benefits against some other attack methods like Deepfool attack.Yayın Closeness and uncertainty aware adversarial examples detection in adversarial machine learning(Elsevier Ltd, 2022-07) Tuna, Ömer Faruk; Çatak, Ferhat Özgür; Eskil, Mustafa TanerWhile deep learning models are thought to be resistant to random perturbations, it has been demonstrated that these architectures are vulnerable to deliberately crafted perturbations, albeit being quasi-imperceptible. These vulnerabilities make it challenging to deploy Deep Neural Network (DNN) models in security-critical areas. Recently, many research studies have been conducted to develop defense techniques enabling more robust models. In this paper, we target detecting adversarial samples by differentiating them from their clean equivalents. We investigate various metrics for detecting adversarial samples. We first leverage moment-based predictive uncertainty estimates of DNN classifiers derived through Monte-Carlo (MC) Dropout Sampling. We also introduce a new method that operates in the subspace of deep features obtained by the model. We verified the effectiveness of our approach on different datasets. Our experiments show that these approaches complement each other, and combined usage of all metrics yields 99 % ROC-AUC adversarial detection score for well-known attack algorithms.Yayın Recovery of impenetrable rough surface profiles via CNN-based deep learning architecture(Taylor and Francis Ltd., 2022-08-18) Aydın, İzde; Budak, Güven; Sefer, Ahmet; Yapar, AliIn this paper, a convolutional neural network (CNN)-based deep learning (DL) architecture for the solution of an electromagnetic inverse problem related to imaging of the shape of the perfectly electric conducting (PEC) rough surfaces is addressed. The rough surface is illuminated by a plane wave and scattered field data is obtained synthetically through the numerical solution of surface integral equations. An effective CNN-DL architecture is implemented through the modelling of the rough surface variation in terms of convenient spline type base functions. The algorithm is numerically tested with various scenarios including amplitude only data and shown that it is very effective and useful.Yayın TENET: a new hybrid network architecture for adversarial defense(Springer Science and Business Media Deutschland GmbH, 2023-08) Tuna, Ömer Faruk; Çatak, Ferhat Özgür; Eskil, Mustafa TanerDeep neural network (DNN) models are widely renowned for their resistance to random perturbations. However, researchers have found out that these models are indeed extremely vulnerable to deliberately crafted and seemingly imperceptible perturbations of the input, referred to as adversarial examples. Adversarial attacks have the potential to substantially compromise the security of DNN-powered systems and posing high risks especially in the areas where security is a top priority. Numerous studies have been conducted in recent years to defend against these attacks and to develop more robust architectures resistant to adversarial threats. In this study, we propose a new architecture and enhance a recently proposed technique by which we can restore adversarial samples back to their original class manifold. We leverage the use of several uncertainty metrics obtained from Monte Carlo dropout (MC Dropout) estimates of the model together with the model’s own loss function and combine them with the use of defensive distillation technique to defend against these attacks. We have experimentally evaluated and verified the efficacy of our approach on MNIST (Digit), MNIST (Fashion) and CIFAR10 datasets. In our experiments, we showed that our proposed method reduces the attack’s success rate lower than 5% without compromising clean accuracy.Yayın Exploiting epistemic uncertainty of the deep learning models to generate adversarial samples(Springer, 2022-03) Tuna, Ömer Faruk; Çatak, Ferhat Özgür; Eskil, Mustafa TanerDeep neural network (DNN) architectures are considered to be robust to random perturbations. Nevertheless, it was shown that they could be severely vulnerable to slight but carefully crafted perturbations of the input, termed as adversarial samples. In recent years, numerous studies have been conducted in this new area called ``Adversarial Machine Learning” to devise new adversarial attacks and to defend against these attacks with more robust DNN architectures. However, most of the current research has concentrated on utilising model loss function to craft adversarial examples or to create robust models. This study explores the usage of quantified epistemic uncertainty obtained from Monte-Carlo Dropout Sampling for adversarial attack purposes by which we perturb the input to the shifted-domain regions where the model has not been trained on. We proposed new attack ideas by exploiting the difficulty of the target model to discriminate between samples drawn from original and shifted versions of the training data distribution by utilizing epistemic uncertainty of the model. Our results show that our proposed hybrid attack approach increases the attack success rates from 82.59% to 85.14%, 82.96% to 90.13% and 89.44% to 91.06% on MNIST Digit, MNIST Fashion and CIFAR-10 datasets, respectively.Yayın Retinal disease diagnosis in OCT scans using a foundational model(Springer Science and Business Media Deutschland GmbH, 2025) Nazlı, Muhammet Serdar; Turkan, Yasemin; Tek, Faik Boray; Toslak, Devrim; Bulut, Mehmet; Arpacı, Fatih; Öcal, Mevlüt CelalThis study examines the feasibility and performance of using single OCT slices from the OCTA-500 dataset to classify DR (Diabetic Retinopathy) and AMD (Age-Related Macular Degeneration) with a pre-trained transformer-based model (RETFound). The experiments revealed the effective adaptation capability of the pretrained model to the retinal disease classification problem. We further explored the impact of using different slices from the OCT volume, assessing the sensitivity of the results to the choice of a single slice (e.g., “middle slice”) and whether analyzing both horizontal and vertical cross-sectional slices could improve outcomes. However, deep neural networks are complex systems that do not indicate directly whether they have learned and generalized the disease appearance as human experts do. The original dataset lacked disease localization annotations. Therefore, we collected new disease classification and localization annotations from independent experts for a subset of OCTA-500 images. We compared RETFound’s explainability-based localization outputs with these newly collected annotations and found that the region attributions aligned well with the expert annotations. Additionally, we assessed the agreement and variability between experts and RETFound in classifying disease conditions. The Kappa values, ranging from 0.35 to 0.69, indicated moderate agreement among experts and between the experts and the model. The transformer-based RETFound model using single or multiple OCT slices, is an efficient approach to diagnosing AMD and DR.
