Makale Koleksiyonu | Bilgisayar Mühendisliği Bölümü

Bu koleksiyon için kalıcı URI

https://hdl.handle.net/11729/4897

Listeleniyor 1 - 20 / 75

A novel similarity based unsupervised technique for training convolutional filters
(IEEE, 2023-05-17) Erkoç, Tuğba; Eskil, Mustata Taner
Achieving satisfactory results with Convolutional Neural Networks (CNNs) depends on how effectively the filters are trained. Conventionally, an appropriate number of filters is carefully selected, the filters are initialized with a proper initialization method and trained with backpropagation over several epochs. This training scheme requires a large labeled dataset, which is costly and time-consuming to obtain. In this study, we propose an unsupervised approach that extracts convolutional filters from a given dataset in a self-organized manner by processing the training set only once without using backpropagation training. The proposed method allows for the extraction of filters from a given dataset in the absence of labels. In contrast to previous studies, we no longer need to select the best number of filters and a suitable filter weight initialization scheme. Applying this method to the MNIST, EMNIST-Digits, Kuzushiji-MNIST, and Fashion-MNIST datasets yields high test performances of 99.19%, 99.39%, 95.03%, and 90.11%, respectively, without applying backpropagation training or using any preprocessed and augmented data.
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 Taner
Deep 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.
Mitosis detection using generic features and an ensemble of cascade adaboosts
(Elsevier, 2013-05-30) Tek, Faik Boray
Context: Mitosis count is one of the factors that pathologists use to assess the risk of metastasis and survival of the patients, which are affected by the breast cancer. Aims: We investigate an application of a set of generic features and an ensemble of cascade adaboosts to the automated mitosis detection. Calculation of the features rely minimally on object -level descriptions and thus require minimal segmentation. Materials and Methods: The proposed work was developed and tested on International Conference on Pattern Recognition (ICPR) 2012 mitosis detection contest data. Statistical Analysis Used: We plotted receiver operating characteristics curves of true positive versus false positive rates; calculated recall, precision, F -measure, and region overlap ratio measures. Results: We tested our features with two different classifier configurations: 1)An ensemble of single adaboosts, 2) an ensemble of cascade adaboosts. On the ICPR 2012 mitosis detection contest evaluation, the cascade ensemble scored 54, 62.7, and 58, whereas the non -cascade version scored 68, 28.1, and 39.7 for the recall, precision, and F -measure measures, respectively. Mostly used features in the adaboost classifier rules were a shape?based feature, which counted granularity and a color-based feature, which relied on Red, Green, and Blue channel statistics. Conclusions: The features, which express the granular structure and color variations, are found useful for mitosis detection. The ensemble of adaboosts performs better than the individual adaboost classifiers. Moreover, the ensemble of cascaded adaboosts was better than the ensemble of single adaboosts for mitosis detection.
A comparison of Auto Train Brain neurofeedback rewarding interfaces in terms of efficacy
(Acıbadem Mehmet Ali Aydınlar Üniversitesi, 2023-01-01) Eroğlu, Günet
Background/aim: Auto Train Brain is a mobile app that was specifically developed for dyslexic children to increase their reading speed and reading comprehension. In the original mobile app, only one unique neurofeedback user interface provided visually and audibly rewarding feedback to the subject with a red-green colored arrow on the screen. Later, new modules are added to the app with the end-users requests. These are the “youtube” video-based interface and “Spotify” auditory-based interface. In this research, we have compared the efficacy of the neurofeedback rewarding interfaces. Materials and methods: The experiment group consists of 20 dyslexic children aged 7-to 10 (15 males, 5 females) who were randomly assigned to one rewarding interface and used it at home for more than six months. Results: The result indicates that though the “youtube” interface is liked most by the participants, the arrow-based simple neurofeedback interface reduces theta brain waves more than other rewarding schemes. On the other hand, “youtube” and “Spotify” based interfaces increase Beta band powers more than the arrow interfaces in the cortex. The ”Spotify” user interface improves the fast brain waves more on the temporal lobes (T7 and T8) as the feedback given was only auditory. Conclusion: The results indicate that the relevant neurofeedback rewarding interface should be chosen based on the dyslexic child’s specific condition.
A novel hybrid edge detection technique: ABC-FA
(ISRES Organizasyon Turizm Eğitim Danışmanlık Ltd. Şti., 2017-11-09) Yelmenoğlu, Elif Deniz; Çelebi, Numan; Taşçı, Tuğrul
Image processing is a vast research field with diversified set of practices utilized in so many application areas such as military, security, medical imaging, machine learning and computer vision based on extracted useful information from any kind of image data. Edges within images are undoubtedly accepted as one of the most significant features providing substantial practical information for various applications working on top of miscellaneous optimization algorithms to achieve better results. Artificial Bee Colony and Firefly algorithms are recently developed optimization algorithms and are used to obtain better results for various problems. In this study, a novel hybrid optimization technique is proposed by combining those algorithms aiming better quality in edge detection on grayscale images. The performance of the proposed algorithm is compared with individual performances of Artificial Bee Colony algorithm and the fundamental edge detection methods. The results are demonstrated that the proposed method is encouraging and also produces meaningful results for similar applications.
Distribution games: a new class of games with application to user provided networks
(Institute of Electrical and Electronics Engineers Inc., 2022-11-29) Taşçı, Sinan Emre; Shalom, Mordechai; Korçak, Ömer
User Provided Network (UPN) is a promising solution for sharing the limited network resources by utilizing user capabilities as a part of the communication infrastructure. In UPNs, it is an important problem to decide how to share the resources among multiple clients in decentralized manner. Motivated by this problem, we introduce a new class of games termed distribution games that can be used to distribute efficiently and fairly the bandwidth capacity among users. We show that every distribution game has at least one pure strategy Nash equilibrium (NE) and any best response dynamics always converges to such an equilibrium. We consider social welfare functions that are weighted sums of bandwidths allocated to clients. We present tight upper bounds for the price of anarchy and price of stability of these games provided that they satisfy some reasonable assumptions. We define two specific practical instances of distribution games that fit these assumptions. We conduct experiments on one of these instances and demonstrate that in most of the settings the social welfare obtained by the best response dynamics is very close to the optimum. Simulations show that this game also leads to a fair distribution of the bandwidth.
Graph convolutional network based virus-human protein-protein interaction prediction for novel viruses
(Elsevier Ltd, 2022-08-13) Koca, Mehmet Burak; Nourani, Esmaeil; Abbasoğlu, Ferda; Karadeniz, İlknur; Sevilgen, Fatih Erdoğan
Computational identification of human-virus protein-protein interactions (PHIs) is a worthwhile step towards understanding infection mechanisms. Analysis of the PHI networks is important for the determination of path-ogenic diseases. Prediction of these interactions is a popular problem since experimental detection of PHIs is both time-consuming and expensive. The available methods use biological features like amino acid sequences, molecular structure, or biological activities for prediction. Recent studies show that the topological properties of proteins in protein-protein interaction (PPI) networks increase the performance of the predictions. The basic network projections, random-walk-based models, or graph neural networks are used for generating topologically enriched (hybrid) protein embeddings. In this study, we propose a three-stage machine learning pipeline that generates and uses hybrid embeddings for PHI prediction. In the first stage, numerical features are extracted from the amino acid sequences using the Doc2Vec and Byte Pair Encoding method. The amino acid embeddings are used as node features while training a modified GraphSAGE model, which is an improved version of the graph convolutional network. Lastly, the hybrid protein embeddings are used for training a binary interaction classifier model that predicts whether there is an interaction between the given two proteins or not. The proposed method is evaluated with comprehensive experiments to test its functionality and compare it with the state-of-art methods. The experimental results on the benchmark dataset prove the efficiency of the proposed model by having a 3–23% better area under curve (AUC) score than its competitors.
Programlamaya giriş dersini alan öğrencilerin programlama öz yeterlilik algılarının ve programlamaya bakış açılarının incelenmesi
(Düzce Üniversitesi, 2021-05-29) Benli, Kristin Surpuhi; Tek, Faik Boray
Bu çalışmada üniversite öğrencilerinin Java programlama öz yeterlilik algıları, programlama öğrenme istekleri ve çalışma alışkanlıkları çeşitli değişkenlere göre (cinsiyet, bölüm, eğitim dili, harf notu, ders tekrarları vb.) istatistiksel yöntemler kullanılarak (T-testi, Mann Whitney U-testi, Kruskal Wallis H testi, tek yönlü varyans analizi, Ki-Kare testi) incelenmiştir. Çalışma grubu, farklı bölümlerde zorunlu olarak programlamaya giriş dersini alan 191 lisans öğrencisinden oluşmaktadır. Elde edilen sonuçlara göre öğrencilerin Java programlama öz yeterlilik algıları bölümlerine ve programlama öğrenme isteklerine göre farklılaşmaktadır. Çalışmada ayrıca Apriori algoritması kullanılarak birliktelik kuralları çıkartılmıştır. En yüksek güven değeri elde edilen kurala göre, programlama öğrenmeyi çok fazla isteyen, programlama öğrenmenin iş hayatında kendisine fayda sağlayacağını düşünen ve programlama dersinden başarı ile geçen öğrencilerin programlama öz yeterlilikleri yüksektir.
A short proof of the size of edge-extremal chordal graphs
(Mahmut Akyiğit, 2022-08-30) Shalom, Mordechai
Blair et. al. [3] have recently determined the maximum number of edges of a chordal graph with a maximum degree less than d and the matching number at most ? by exhibiting a family of chordal graphs achieving this bound. We provide simple proof of their result.
From past to present: spam detection and identifying opinion leaders in social networks
(Yildiz Teknik Univ., 2022-06-22) Altınel Girgin, Ayşe Berna; Gümüşçekiçci, Gizem
On microblogging sites, which are gaining more and more users every day, a wide range of ideas are quickly emerging, spreading, and creating interactive environments. In some cases, in Turkey as well as in the rest of the world, it was noticed that events were published on microblogging sites before appearing in visual, audio and printed news sources. Thanks to the rapid flow of information in social networks, it can reach millions of people in seconds. In this context, social media can be seen as one of the most important sources of information affecting public opinion. Since the information in social networks became accessible, research started to be conducted using the information on the social networks. While the studies about spam detection and identification of opinion leaders gained popularity, surveys about these topics began to be published. This study also shows the importance of spam detection and identification of opinion leaders in social networks. It is seen that the data collected from social platforms, especially in recent years, has sourced many state-of-art applications. There are independent surveys that focus on filtering the spam content and detecting influencers on social networks. This survey analyzes both spam detection studies and opinion leader identification and categorizes these studies by their methodologies. As far as we know there is no survey that contains approaches for both spam detection and opinion leader identification in social networks. This survey contains an overview of the past and recent advances in both spam detection and opinion leader identification studies in social networks. Furthermore, readers of this survey have the opportunity of understanding general aspects of different studies about spam detection and opinion leader identification while observing key points and comparisons of these studies.
ComStreamClust: a communicative multi-agent approach to text clustering in streaming data
(Springer Science and Business Media Deutschland GmbH, 2023-12) Najafi, Ali; Gholipour-Shilabin, Araz; Dehkharghani, Rahim; Mohammadpur-Fard, Ali; Asgari-Chenaghlu, Meysam
Topic detection is the task of determining and tracking hot topics in social media. Twitter is arguably the most popular platform for people to share their ideas with others about different issues. One such prevalent issue is the COVID-19 pandemic. Detecting and tracking topics on these kinds of issues would help governments and healthcare companies deal with this phenomenon. In this paper, we propose a novel, multi-agent, communicative clustering approach, so-called ComStreamClust for clustering sub-topics inside a broader topic, e.g., the COVID-19 and the FA CUP. The proposed approach is parallelizable, and can simultaneously handle several data-point. The LaBSE sentence embedding is used to measure the semantic similarity between two tweets. ComStreamClust has been evaluated by several metrics such as keyword precision, keyword recall, and topic recall. Based on topic recall on different number of keywords, ComStreamClust obtains superior results when compared to the existing methods.
Closeness and uncertainty aware adversarial examples detection in adversarial machine learning
(Elsevier Ltd, 2022-07) Tuna, Ömer Faruk; Çatak, Ferhat Özgür; Eskil, Mustafa Taner
While 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.
Battle Royale Optimizer for solving binary optimization problems
(Elsevier B.V., 2022-05) Akan, Taymaz; Agahian, Saeid; Dehkharghani, Rahim
Battle Royale Optimizer (BRO) is a recently proposed metaheuristic optimization algorithm used only in continuous problem spaces. The BinBRO is a binary version of BRO. The BinBRO algorithm employs a differential expression, which utilizes a dissimilarity measure between binary vectors instead of a vector subtraction operator, used in the original BRO algorithm to find the nearest neighbor. To evaluate BinBRO, we applied it to two popular benchmark datasets: the uncapacitated facility location problem (UFLP) and the maximum-cut (Max-Cut) graph problems from OR-Library. An open-source MATLAB implementation of BinBRO is available on CodeOcean and GitHub websites.
Uncertainty as a Swiss army knife: new adversarial attack and defense ideas based on epistemic uncertainty
(Springer, 2022-04-02) Tuna, Ömer Faruk; Çatak, Ferhat Özgür; Eskil, Mustafa Taner
Although state-of-the-art deep neural network models are known to be robust to random perturbations, it was verified that these architectures are indeed quite vulnerable to deliberately crafted perturbations, albeit being quasi-imperceptible. These vulnerabilities make it challenging to deploy deep neural network models in the areas where security is a critical concern. In recent years, many research studies have been conducted to develop new attack methods and come up with new defense techniques that enable more robust and reliable models. In this study, we use the quantified epistemic uncertainty obtained from the model's final probability outputs, along with the model's own loss function, to generate more effective adversarial samples. And we propose a novel defense approach against attacks like Deepfool which result in adversarial samples located near the model's decision boundary. We have verified the effectiveness of our attack method on MNIST (Digit), MNIST (Fashion) and CIFAR-10 datasets. In our experiments, we showed that our proposed uncertainty-based reversal method achieved a worst case success rate of around 95% without compromising clean accuracy.
k-Means clustering by using the calculated Z-scores from QEEG data of children with dyslexia
(Taylor & Francis, 2023) Eroğlu, Günet; Arman, Fehim
Learning the subtype of dyslexia may help shorten the rehabilitation process and focus more on the relevant special education or diet for children with dyslexia. For this purpose, the resting-state eyes-open 2-min QEEG measurement data were collected from 112 children with dyslexia (84 male, 28 female) between 7 and 11 years old for 96 sessions per subject on average. The z-scores are calculated for each band power and each channel, and outliers are eliminated afterward. Using the k-Means clustering method, three different clusters are identified. Cluster 1 (19% of the cases) has positive z-scores for theta, alpha, beta-1, beta-2, and gamma-band powers in all channels. Cluster 2 (76% of the cases) has negative z-scores for theta, alpha, beta-1, beta-2, and gamma-band powers in all channels. Cluster 3 (5% of the cases) has positive z-scores for theta, alpha, beta-1, beta-2, and gamma-band powers at AF3, F3, FC5, and T7 channels and mostly negative z-scores for other channels. In Cluster 3, there is temporal disruption which is a typical description of dyslexia. In Cluster 1, there is a general brain inflammation as both slow and fast waves are detected in the same channels. In Cluster 2, there is a brain maturation delay and a mild inflammation. After Auto Train Brain training, most of the cases resemble more of Cluster 2, which may mean that inflammation is reduced and brain maturation delay comes up to the surface which might be the result of inflammation. Moreover, Cluster 2 center values at the posterior parts of the brain shift toward the mean values at these channels after 60 sessions. It means, Auto Train Brain training improves the posterior parts of the brain for children with dyslexia, which were the most relevant regions to be strengthened for dyslexia.
Exploiting epistemic uncertainty of the deep learning models to generate adversarial samples
(Springer, 2022-03) Tuna, Ömer Faruk; Çatak, Ferhat Özgür; Eskil, Mustafa Taner
Deep 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.
BinBRO: Binary Battle Royale Optimizer algorithm
(Elsevier Ltd, 2022-02-04) (Rahkar Farshi), Taymaz Akan; Agahian, Saeid; Dehkharghani, Rahim
Stochastic methods attempt to solve problems that cannot be solved by deterministic methods with reasonable time complexity. Optimization algorithms benefit from stochastic methods; however, they do not guarantee to obtain the optimal solution. Many optimization algorithms have been proposed for solving problems with continuous nature; nevertheless, they are unable to solve discrete or binary problems. Adaptation and use of continuous optimization algorithms for solving discrete problems have gained growing popularity in recent decades. In this paper, the binary version of a recently proposed optimization algorithm, Battle Royale Optimization, which we named BinBRO, has been proposed. The proposed algorithm has been applied to two benchmark datasets: the uncapacitated facility location problem, and the maximum-cut graph problem, and has been compared with 6 other binary optimization algorithms, namely, Particle Swarm Optimization, different versions of Genetic Algorithm, and different versions of Artificial Bee Colony algorithm. The BinBRO-based algorithms could rank first among those algorithms when applying on all benchmark datasets of both problems, UFLP and Max-Cut.
Evaluating the English-Turkish parallel treebank for machine translation
(TÜBİTAK, 2022-01-19) Görgün, Onur; Yıldız, Olcay Taner
This study extends our initial efforts in building an English-Turkish parallel treebank corpus for statistical machine translation tasks. We manually generated parallel trees for about 17K sentences selected from the Penn Treebank corpus. English sentences vary in length: 15 to 50 tokens including punctuation. We constrained the translation of trees by (i) reordering of leaf nodes based on suffixation rules in Turkish, and (ii) gloss replacement. We aim to mimic human annotator's behavior in real translation task. In order to fill the morphological and syntactic gap between languages, we do morphological annotation and disambiguation. We also apply our heuristics by creating Nokia English-Turkish Treebank (NTB) to address technical document translation tasks. NTB also includes 8.3K sentences in varying lengths. We validate the corpus both extrinsically and intrinsically, and report our evaluation results regarding perplexity analysis and translation task results. Results prove that our heuristics yield promising results in terms of perplexity and are suitable for translation tasks in terms of BLEU scores.
On the online coalition structure generation problem
(AI Access Foundationusc Information Sciences Inst, 2021) Flammini, Michele; Monaco, Gianpiero; Moscardelli, Luca; Shalom, Mordechai; Zaks, Shmuel
We consider the online version of the coalition structure generation problem, in which agents, corresponding to the vertices of a graph, appear in an online fashion and have to be partitioned into coalitions by an authority (i.e., an online algorithm). When an agent appears, the algorithm has to decide whether to put the agent into an existing coalition or to create a new one containing, at this moment, only her. The decision is irrevocable. The objective is partitioning agents into coalitions so as to maximize the resulting social welfare that is the sum of all coalition values. We consider two cases for the value of a coalition: (1) the sum of the weights of its edges, and (2) the sum of the weights of its edges divided by its size. Coalition structures appear in a variety of application in AI, multi-agent systems, networks, as well as in social networks, data analysis, computational biology, game theory, and scheduling. For each of the coalition value functions we consider the bounded and unbounded cases depending on whether or not the size of a coalition can exceed a given value alpha. Furthermore, we consider the case of a limited number of coalitions and various weight functions for the edges, i.e., unrestricted, positive and constant weights. We show tight or nearly tight bounds for the competitive ratio in each case.
On the maximum cardinality cut problem in proper interval graphs and related graph classes
(Elsevier B.V., 2022-01-04) Boyacı, Arman; Ekim, Tınaz; Shalom, Mordechai
Although it has been claimed in two different papers that the maximum cardinality cut problem is polynomial-time solvable for proper interval graphs, both of them turned out to be erroneous. In this work we consider the parameterized complexity of this problem. We show that the maximum cardinality cut problem in proper/unit interval graphs is FPT when parameterized by the maximum number of non-empty bubbles in a column of its bubble model. We then generalize this result to a more general graph class by defining new parameters related to the well-known clique-width parameter. Specifically, we define an (?,?,?)-clique-width decomposition of a graph as a clique-width decomposition in which at each step the following invariant is preserved: after discarding at most ? labels, a) every label consists of at most ? sets of twin vertices, and b) all the labels together induce a graph with independence number at most ?. We show that for every two constants ?,?>0 the problem is FPT when parameterized by ? plus the smallest width of an (?,?,?)-clique-width decomposition.

Güncel Gönderiler