Yazar "Kabakuş, Abdullah Talha" seçeneğine göre listele
Listeleniyor 1 - 20 / 27
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe A Performance Comparison of SQLite and Firebase Databases from A Practical Perspective(Düzce Üniversitesi, 2019) Kabakuş, Abdullah TalhaAndroid is currently the most used mobile operating system all over the world. The two database management systems that Android officially supports are SQLite and Firebase. Android SDK provides built-in packages to let developers implement applications which store its data on these databases. At this point, it is necessary to reveal the performance comparison of these databases. For this reason, an Android application that evaluates several experiments which cover the most used data operations on these databases is implemented within this study. The experimental result indicates that SQLite provides better performance compared to Firebase except deleting data. The performance differences between SQLite and Firebase vary through (1) the type of data operation, and (2) the size of data that is managed.Öğe A Comparison of the State-of-the-Art Deep Learning Platforms: An Experimental Study(2020) Kabakuş, Abdullah TalhaDeep learning, a subfield of machine learning, has proved its efficacy on a wide range of applications includingbut not limited to computer vision, text analysis and natural language processing, algorithm enhancement,computational biology, physical sciences, and medical diagnostics by producing results superior to the state-ofthe-art approaches. When it comes to the implementation of deep neural networks, there exist various state-of-theart platforms. Starting from this point of view, a qualitative and quantitative comparison of the state-of-the-artdeep learning platforms is proposed in this study in order to shed light on which platform should be utilized forthe implementations of deep neural networks. Two state-of-the-art deep learning platforms, namely, () Keras, and() PyTorch were included in the comparison within this study. The deep learning platforms were quantitativelyexamined through the models based on three most popular deep neural networks, namely, () Feedforward NeuralNetwork (FNN), () Convolutional Neural Network (CNN), and () Recurrent Neural Network (RNN). Themodels were evaluated on three evaluation metrics, namely, () training time, () testing time, and () predictionaccuracy. According to the experimental results, while Keras provided the best performance for both FNNs andCNNs, PyTorch provided the best performance for RNNs expect for one evaluation metric, which was the testingtime. This experimental study should help deep learning engineers and researchers to choose the most suitableplatform for the implementations of their deep neural networks.Öğe The Data Science Met with the COVID-19: Revealing the Most Critical Measures Taken for the COVID-19 Pandemic(2020) Kabakuş, Abdullah TalhaThe whole world has been fighting against the novel coronavirus 2019 (COVID-19) for months. Despite theadvances in medical sciences, more than 235,000 people have died so far. And, despite all the measures taken forit, more than 3 million people have become sick of the COVID-19. The measures taken for the COVID-19 varythrough countries. So, revealing the most critical measures is necessary for a better fight against both the COVID19 and possible similar pandemics in the future. To this end, an analysis of the worldwide measures, which weretaken so far, for the COVID-19 pandemic was proposed within this paper. Since it is still early days, for the bestof our knowledge, there does not exist a single dataset contains all the features utilized within this study. Therefore,a novel global dataset containing the data regarding the COVID-19 for 52 countries around the world wasconstructed by combining various datasets. Then, the feature importance techniques were employed to reveal theimportance of the utilized features which means revealing the most important measures taken for the COVID-19pandemic for our case. Within the analysis, four features were utilized, namely, the population density, the walkingmobility, the driving mobility, and the number of lockdown days. According to the experimental result, thepopulation density was found as the most important feature which means the most critical measure in terms ofincreasing the spread of the COVID-19 pandemic. The order of the importance of the other features was found asthe walking mobility, the driving mobility, and the number of lockdown days, respectively.Öğe ddosdaps4web: Web'e Yönelik DDoS Tespit ve Koruma Yöntemi(2016) Kabakuş, Abdullah Talha; Kara, ResulHer koruma tespitle başlar. Dağıtık servis engelleme (DDoS) saldırıları, ağları veya bilgisayarlara yoğunkullanım sonucunda verdikleri servisi engellenmektedirler. Günümüzde bilgisayarlardaki yazılımsal vedonanımsal gelişmelere rağmen, kısa bir zaman dilimi DDoS ataklarının kötücül etkilerini gerçekleştirmesiiçin yeterli olmaktadır. Bu sebepten ötürü DDos saldırılarını engellemek için gerçek zamanlı bir tespit vekoruma sistemine ihtiyaç duyulmaktadır. Geleneksel ağ tabanlı koruma sistemleri uygulama katmanı DDoSataklarına karşı güvenlik sağlayamamaktadır. Bu çalışmada, HTTP tabanlı DDoS ataklarını tespit etmek vesistemi korumak için ddosdaps4web isimli DDoS tespit ve koruma sistemi öne sürülmüştür. ddosdaps4webüç servisten faydalanmaktadır: (1) Tüm HTTP isteklerinin depolanıp, istek başlıklarından detaylı analiz içinbilgi çıkartımı yapılmasını sağlayan depolama servisi, (2) her dakika çalışan ve ön tanımlı istek limitlerinegöre kötücül istekleri tespit etmeyi sağlayan izleme servisi, ve (3) gelen bütün istekleri keserek, oluşturulankurallara göre kötücül olanları devre dışı bırakan durdurucu servisi. ddosdaps4web rastgele oluşturulmuş10000 HTTP isteği üzerinden test edilerek DDoS doğru tespit oranı %94 olarak bulunmuştur.Öğe DroidMalwareDetector: A novel Android malware detection framework based on convolutional neural network(Pergamon-Elsevier Science Ltd, 2022) Kabakuş, Abdullah TalhaSmartphones have become an integral part of our daily lives thanks to numerous reasons. While benefitting from what they offer, it is critical to be aware of the existence of malware in the Android ecosystem and be away from them. To this end, an end-to-end and highly effective Android malware detection framework based on CNN, namely, DroidMalwareDetector, was proposed within this study. Unlike most of the related work, DroidMalwar-eDetector was specifically designed to (i) automate feature extraction and selection, (ii) propose a novel CNN that operates with 1-dimensional data, and (iii) use intents and API calls alongside the widely used permissions to perform comprehensive malware analysis. The proposed framework was trained and evaluated on the con-structed dataset, which consisted of 14,386 apps from the de-facto standard datasets. The proposed framework's efficiency in terms of distinguishing malware from benign apps was revealed thanks to the conducted experi-ments. According to the experimental result, the accuracy of the proposed framework was calculated as high as 0.9, which was higher than the accuracy values obtained from a wide range of machine learning algorithms. The insights which were gained through the conducted experiments were revealed as another contribution to the research field.Öğe The Effect of Data and Object Types on Java Virtual Machine(2018) Kabakuş, Abdullah TalhaHow the data is stored in the memory becomes more critical when the size of data increases. The programming languages define data and object types that can be used while programming. Most programming languages provide more than one data and object type in order to let developers use more sensitive types which address their needs. Memory management is a key concept for the data-intensive systems. Also, the NoSQL databases, which are alternatives to relational database management systems, tend to store the data in memory and serve the data from memory. In this study, the effect of data and object types on Java Virtual Machine is evaluated in order to reveal its effect in terms of consumed memory on Java programming language. Experimental results reveal some key points for developers to use memory more efficiently.Öğe EMPOWERING SELF-DETECTION: A GRAPHICAL USER INTERFACE POWERED BY MACHINE LEARNING FOR EARLY DIAGNOSIS OF ALZHEIMER'S DISEASE(Istanbul Ticaret University, 2024) Kabakuş, Abdullah Talha; Erdoğmuş, PakizeAlzheimer’s Disease (AD) is one of the most, if not the most, devastating neurodegenerative diseases that are incurable and progressive. Early diagnosis of AD comes with many promises in terms of medicine, sociology, and economics. Despite the existence of numerous studies that aim for early diagnosis of AD, to the best of our knowledge, there is not a publicly available tool that lets end-users assess AD. To address this gap, we propose a Graphical User Interface (GUI) powered by Machine Learning (ML) that makes self-assessment of AD possible – without any input from medical experts. The developed GUI lets end-users enter various information considering both commonly used features for the diagnosis of AD and the questions available in the gold standard screening tool for the diagnosis of AD, namely the Mini-Mental State Exam. In addition to employing 11 traditional ML algorithms, we propose a novel 1-dimensional (1D) Convolutional Neural Network (CNN). All ML models were trained on a gold standard dataset that comprised 373 records from three subject classes as follows: (i) non-demented, (ii) demented, and (iii) converted. Once the end-user enters the required input through the developed GUI, the previously trained ML model assesses the diagnosis of AD through this input in a couple of seconds. According to the experimental results, the proposed novel 1D CNN outperformed the state-of-the-art by obtaining an accuracy as high as 95,3% on the used gold standard dataset.Öğe An experimental comparison of the widely used pre-trained deep neural networks for image classification tasks towards revealing the promise of transfer-learning(Wiley, 2022) Kabakuş, Abdullah Talha; Erdogmus, PakizeThe easiest way to propose a solution based on deep neural networks is using the pre-trained models through the transfer-learning technique. Deep learning platforms provide various pre-trained deep neural networks that can be easily applied for image classification tasks. So, Which pre-trained model provides the best performance for image classification tasks? is a question that instinctively comes to mind and should be shed light on by the research community. To this end, we propose an experimental comparison of the six popular pre-trained deep neural networks, namely, (i) VGG19, (ii) ResNet50, (iii) DenseNet201, (iv) MobileNetV2, (v) InceptionV3, and (vi) Xception by employing them through the transfer-learning technique. Then, the proposed benchmark models were both trained and evaluated under the same configurations on two gold-standard datasets, namely, (i) CIFAR-10 and (ii) Stanford Dogs to benchmark them. Three evaluation metrics were employed to measure performance differences between the employed pre-trained models as follows: (i) Accuracy, (ii) training duration, and (iii) inference time. The key findings that were obtained through the conducted a wide variety of experiments were discussed.Öğe Finding Influencers on Twitter with Using Machine Learning Classification Algorithms(Aydın KARAPINAR, 2018) Şimşek, Mehmet; Kabakuş, Abdullah TalhaMicroblog sites are environments where people follow people. With this feature, a microblog site is a convenient environment for spreading an opinion or introducing a new product. The key point is determination of individuals who maximize the spreading. This problem is known as Influence Maximization (IM) and has attracted attention of many researchers. Many studies in the literature have modeled IM problem on graphs for different propagation models such as Independent Cascade (IC) and Linear Threshold (LT). However, microblogs like Twitter have their own features. Many works on IM in Twitter derive new metrics from user and tweet features; apply a greedy approach for selecting influencers. In this study, we adopted different approach for IM problem, and we dealt it as a classification problem. Firstly, we collected data on International Women Day 2018; empirically we labeled the users as either influencer candidates or non-influencers; then we applied classification methods for classifying users into one class with using features of users. By this way, we obtained an influencer candidates set, which is very smaller than entire dataset. Experimental results show that making selection with using same heuristic (namely MF) from the reduced influencer candidates set outperforms making selection from entire dataset.Öğe GRAPH-BASED SENTENCE LEVEL SPELL CHECKING FRAMEWORK(Inst Integrative Omics & Applied Biotechnology, 2017) Kabakuş, Abdullah Talha; Kara, ResulSpelling mistakes are very common on the web, especially when it comes to social media, it is much more common since (1) users tend to use an informal language that contains slang, and (2) the character limit defined by some social services such as Twitter. Traditional string similarity measurements (1) do not consider the context of the misspelled word while providing alternatives, and (2) do not provide a certain way to choose the right word when there are multiple alternatives that have the same similarity with the misspelled word. Therefore, we propose a novel sentence level spell checking framework that targets to find "the most frequently used similar alternative word". 146,808 sentences from different corpora are stored in a graph database. The similarity is calculated by using Levenshtein distance algorithm alongside the similarity between two given words. As the experimental results are presented in the discussion, the proposed framework is able to correct misspellings which cannot be corrected by traditional string similarity measurement based approaches. The accuracy of the proposed framework is calculated as 84%. Since the proposed framework uses a slang dictionary to determine misspelled words, it can be used to correct misspellings in the social media platforms.Öğe Hybroid: A Novel Hybrid Android Malware Detection Framework(2021) Kabakuş, Abdullah TalhaAndroid, the most widely-used mobile operating system, attracts the attention of malware developers as well as benign users. Despite the serious proactive actions taken by Android, the Android malware is still widespread as a result of the increasing sophistication and the diversity of malware. Android malware detection systems are generally classified into two: (1) Static analysis, and (2) dynamic analysis. In this study, a novel Android malware detection framework, namely, Hybroid, was proposed which combines both the static and dynamic analysis techniques to benefit from the advantages of both of these techniques. An up-todate version of Android, namely, Android Oreo, was specifically employed in order to handle the problem from an up-to-date perspective as the recent versions of Android provide new security mechanisms, which are discussed with this study. Hybroid was evaluated on a large dataset that consists of applications, and the accuracy of Hybroid was calculated as high as when it was utilized with the J48 classification algorithm which outperforms the state-of-the-art studies. The key findings in consequence of the experimental result are discussed in order to shed light on Android malware detection.Öğe Öğe An in-depth analysis of Android malware using hybrid techniques(Elsevier Sci Ltd, 2018) Kabakuş, Abdullah Talha; Doğru, İbrahim AlperAndroid malware is widespread despite the effort provided by Google in order to prevent it from the official application market, Play Store. Two techniques namely static and dynamic analysis are commonly used to detect malicious applications in Android ecosystem. Both of these techniques have their own advantages and disadvantages. In this paper, we propose a novel hybrid Android malware analysis approach namely mad4a which uses the advantages of both static and dynamic analysis techniques. The aim of this study is revealing some unknown characteristics of Android malware through the used various analysis techniques. As the result of static and dynamic analysis on the widely used Android application datasets, digital investigators are informed about some underestimated characteristics of Android malware. (c) 2018 Elsevier Ltd. All rights reserved.Öğe NATURAL LANGUAGE QUESTION ANSWERING SYSTEM OVER LINKED DATA(2019) Kabakuş, Abdullah Talha; Çetin, AydınLinked Data project is aimed to give more details on any subject through the big knowledge bases defined on the web. In thiscontext, knowledge bases offer endpoint service user interfaces to query their data. Because of the SPARQL query languagelimitation of these knowledge bases, a significant number of web users are unable to benefit from these services. In this paper,an English natural language question answering system over Linked Data is proposed in order to eliminate this limitation. Theproposed system's main processes can be listed as follows: (1) Extracting Part-Of-Speech (POS) tags, (2) pattern extraction &preparing appropriate SPARQL queries, (3) executing user queries & displaying the results. The features which are notprovided by the endpoint services of knowledge bases such as dynamic paging, voice search and answer vocalization whichmake the usage of the proposed system to be possible by the visually-impaired web users, question-answer caching, socialmedia integration, and live spell checking are proposed. According to experimental results, the proposed system’s questionanswering performance is improved between 2 and 12 times through the type of natural language question thanks to thequestion-answer caching mechanism.Öğe A novel COVID-19 sentiment analysis in Turkish based on the combination of convolutional neural network and bidirectional long-short term memory on Twitter(Wiley, 2022) Kabakuş, Abdullah TalhaThe whole world has been experiencing the COVID-19 pandemic since December 2019. During the pandemic, a new life has been started by necessity where people have extensively used social media to express their feelings, and find information. Twitter was used as the source of what people have shared regarding the COVID-19 pandemic. Sentiment analysis deals with the extraction of the sentiment of a given text. Most of the related works deal with sentiment analysis in English, while studies for Turkish sentiment analysis lack in the research field. To this end, a novel sentiment analysis model based on the combination of convolutional neural network and bidirectional long short-term memory was proposed in this study. The proposed deep neural network model was trained on the constructed Twitter dataset, which consists of 15k Turkish tweets regarding the COVID-19 pandemic, to classify a given tweet into three sentiment classes, namely, (i) positive, (ii) negative, and (iii) neutral. A set of experiments were conducted for the evaluation of the proposed model. According to the experimental result, the proposed model obtained an accuracy as high as 97.895%, which outperformed the state-of-the-art baseline models for sentiment analysis of tweets in Turkish.Öğe A Novel Gender Classification Model based on Convolutional Neural Network through Handwritten Text and Numeral(2023) Erdoğmuş, Pakize; Kabakuş, Abdullah Talha; Küçükkülahlı, Enver; Takgil, Büşra; Kara Timuçin, EzgiHuman handwriting is used to investigate human characteristics in various applications, including but not limited to biometric authentication, personality profiling, historical document analysis, and forensic investigations. Gender is one of the most distinguishing characteristics of human beings. From this point forth, we propose a novel end-to-end model based on Convolutional Neural Network (CNN) that automatically extracts features from a given handwritten sample, which contains both handwritten text and numerals unlike the related work that uses only handwritten text, and classifies its owner’s gender. In addition to proposing a novel model, we introduce a new dataset that consists of 530 gender-labeled Turkish handwritten samples since, to the best of our knowledge, there does not exist a public gender-labeled Turkish handwriting dataset. Following an exhaustive process of hyperparameter optimization, the proposed CNN featured the most optimal hyperparameters and was both trained and evaluated on this dataset. According to the experimental result, the proposed novel model obtained an accuracy as high as 74.46%, which overperformed the state-of-the-art baselines and is promising on such a task that even humans could not have achieved highly-accurate results for, as of yet.Öğe A novel robust convolutional neural network for uniform resource locator classification from the view of cyber security(Wiley, 2023) Kabakuş, Abdullah TalhaUniform resource locator (URL)-based cyber-attacks form a major part of security threats in cyberspace. Even though the experience and awareness of the end-users help them protect themselves from these attacks, a software-based solution is necessary for comprehensive protection. To this end, a novel robust URL classification model based on convolutional neural network is proposed in this study. The proposed model classifies given URLs into five classes, namely, (i$$ \mathrm{i} $$) benign$$ \mathrm{benign} $$, (ii$$ \mathrm{ii} $$) defacement$$ \mathrm{defacement} $$, (iii$$ \mathrm{iii} $$) phishing$$ \mathrm{phishing} $$, (iv$$ \mathrm{iv} $$) spam$$ \mathrm{spam} $$, and (v$$ \mathrm{v} $$) malware$$ \mathrm{malware} $$. The proposed model was trained and evaluated on a gold standard URL dataset comprising of 36,707$$ \mathrm{36,707} $$ samples. According to the experimental result, the proposed model obtained an accuracy as high as 98.1%$$ 98.1\% $$ which outperformed the state-of-the-art. Based on the same architecture, we proposed another classifier, a binary classifier that detects malicious URLs without dealing with their types. This binary classifier obtained an accuracy as high as 99.3%$$ 99.3\% $$ which outperformed the state-of-the-art as well. The experimental result demonstrates the feasibility of the proposed solution.Öğe A Performance Comparison of Java Cache Memory Implementations(2020) Kabakuş, Abdullah TalhaNowadays the information systems are substantially data-intensive and the data is going to be more critical than before.For these systems, which are intolerant in terms of time latency, the way of accessing data becomes more critical. Inthese situations, an additional data layer named cache memory is used. There are various both open-source andcommercial Java cache memory implementations based on the specifications defined by Java Community Process. Inthis study, the most widely used Java cache memory implementations are evaluated in order to compare theirperformances in terms of elapsed time and memory consumption. The experimental results imply that the architecturaldesign of cache memory has a great effect on performance and there is no winner that provides the best performance forall data operations.Öğe A performance evaluation of in-memory databases(Elsevier Science Bv, 2017) Kabakuş, Abdullah Talha; Kara, ResulThe popularity of NoSQL databases has increased due to the need of (1) processing vast amount of data faster than the relational database management systems by taking the advantage of highly scalable architecture, (2) flexible (schema-free) data structure, and, (3) low latency and high performance. Despite that memory usage is not major criteria to evaluate performance of algorithms, since these databases serve the data from memory, their memory usages are also experimented alongside the time taken to complete each operation in the paper to reveal which one uses the memory most efficiently. Currently there exists over 225 NoSQL databases that provide different features and characteristics. So it is necessary to reveal which one provides better performance for different data operations. In this paper, we experiment the widely used in-memory databases to measure their performance in terms of (1) the time taken to complete operations, and (2) how efficiently they use memory during operations. As per the results reported in this paper, there is no database that provides the best performance for all data operations. It is also proved that even though a RDMS stores its data in memory, its overall performance is worse than NoSQL databases. (C) 2016 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University.Öğe Survey of Instant Messaging Applications Encryption Methods(2015) Kabakuş, Abdullah Talha; Kara, ResulAnlık mesajlaşma uygulamaları, kolay kullanımları ve popülaritelerinden dolayı geleneksel Kısa Mesajlaşma Servisi (SMS) ve Çoklu Medya Mesajlaşma Servisi (MMS)'in yerini aldı. Anlık mesajlaşma uygulama kullanıcıları, bu uygulamalar aracılığıyla metin, ses mesajları, fotoğraf, video, kişi bilgisi gibi çeşitli türlerdeki ekleri arkadaşlarıyla gerçek zamanlı olarak paylaşabilmektedir. Anlık mesajlaşma uygulamaları Kısa Mesaj Servisi Teknik Gerçeklemesi (GSM) yerine sadece günümüzde en çok kullanılan iletişim aracı olan internete ihtiyaç duyduğundan dolayı ücretsizdir. Buradaki kritik nokta, siber saldırganlarına ve bilgisayar korsanlarına karşı herhangi açık nokta bırakmamak için bu mesajların güvenliğinin sağlanmasıdır. PricewaterhouseCoopers tarafından yapılan son rapora göre, 2014 yılında tespit edilen uluslararası siber saldırılar sayısı 42.8 milyona çıkarak %48'e yükselmiştir. Postini güvenlik şirketi tarafından yayınlanan başka bir rapor ise anlık mesajlaşmayı hedefleyen tehditlerin %90'ının oldukça yıkıcı soluncanlar olduğunu belirtmektedir. Bu çalışmada, anlık mesajlaşma uygulamalarının şifreleme yöntemleri karşılaştırmalı olarak sunulmuştur. Anlık mesajlaşma uygulamaları üç farklı platform göz önüne alınarak incelenmiştir: (1) Masaüstü istemcileri, (2) web istemcileri ve (3) mobil telefon istemcileri. Anlık mesajlaşma uygulamaları, birçok araştırmada en çok üzerinde durulan kritik kriterler olan (1) internet üzerinden metin dönüşümü, (2) şifreleme sonrası metin dönüşümü ve (3) Güvenli Giriş Katmanı (SSL) kullanıldıktan sonra yapılan metin dönüşümüne göre karşılaştırılmıştır. Son olarak yazarlar, güvenli bir mesajlaşma uygulamasında bulunması gereken kritik gereksinimleri vurgulamıştır
