Yönetim Bilişim Sistemleri Bölümü Koleksiyonu
Permanent URI for this collectionhttps://hdl.handle.net/20.500.12469/68
Browse
Browsing Yönetim Bilişim Sistemleri Bölümü Koleksiyonu by Institution Author "Çayır, Aykut"
Now showing 1 - 5 of 5
- Results Per Page
- Sort Options
Conference Object Citation - WoS: 54Citation - Scopus: 97Bitcoin Forecasting Using Arima and Prophet(IEEE, 2018) Yenidoğan, Işıl; Çayır, Aykut; Kozan, Ozan; Dağ, Tugce; Arslan, ÇiğdemThis paper presents all studies methodology and results about Bitcoin forecasting with PROPHET and ARIMA methods using R analytics platform. To find the most accurate forecast model the performance metrics of PROPHET and AMNIA methods are compared on the same dataset. The dataset selected 16r this study starts from May 2016 and ends in March 2018 which is the interval that Bitcoin values changing significantly against the other currencies. Data is prepared for time series analysis by performing data preprocessing steps such as time stamp conversion and feature selection. Although the time series analysis has a univariate characteristics it is aimed to include some additional variables to each model to improve the forecasting accuracy. Those additional variables are selected based on different correlation studies between cryptocurrencies and real currencies. The model selection for both ARIMA and PROPHET is done by using threefold splitting technique considering the time series characteristics of the dataset. The threefold splitting technique gave the optimum ratios for training validation and test sets. Filially two different models are created and compared in terms of performance metrics. Based on the extensive testing we see that PROPHET outperforms ARIMA by 0.94 to 0.68 in R-2 values.Conference Object Citation - WoS: 39Citation - Scopus: 55Feature Extraction Based on Deep Learning for Some Traditional Machine Learning Methods(Institute of Electrical and Electronics Engineers Inc., 2018) Çayır, Aykut; Yenidoğan, Işıl; Dağ, HasanDeep learning is a subfield of machine learning and deep neural architectures can extract high level features automatically without handcraft feature engineering unlike traditional machine learning algorithms. In this paper, we propose a method, which combines feature extraction layers of a convolutional neural network with traditional machine learning algorithms, such as, support vector machine, gradient boosting machines, and random forest. All of the proposed hybrid models and the above mentioned machine learning algorithms are trained on three different datasets: MNIST, Fashion-MNIST, and CIFAR-10. Results show that the proposed hybrid models are more successful than traditional models while they are being trained from raw pixel values. In this study, we empower traditional machine learning algorithms for classification using feature extraction ability of deep neural network architectures and we are inspired by transfer learning methodology to this.Article Konutların Günlük Elektrik Güç Tüketimi Tahmini İçin Uygun Model Seçimi(Fırat Üniv. Fen Bilimleri Enst., 2018) Çayır, Aykut; Dağ, HasanZamana bağlı değişim gösteren olayların modellenmesi zorlu bir veri analizi problemidir. Bu olaylardan biri olan elektrik güç tüketiminde ise veriden mevsimsel etki ve tatil günleri gibi örüntülerin öğrenilerek bir tüketim tahmin modelinin geliştirilebilmesi için klasik makine öğrenmesi ve derin öğrenme yöntemlerinden yararlanılmaktadır. Bu çalışmada, İngiltere’nin Londra şehrindeki belirli bir bölgede 30 farklı eve ait yaklaşık 3 yıllık elektrik güç tüketimi veri kümesi kullanılarak uygun bir kısa vadeli tüketim tahmin modelinin makine öğrenmesi algoritmaları ile bulunması amaçlanmıştır.Article Citation - Scopus: 48Random Capsnet Forest Model for Imbalanced Malware Type Classification Task(Elsevier, 2021) Çayır, Aykut; Ünal, Uğur; Dağ, HasanBehavior of malware varies depending the malware types, which affects the strategies of the system protection software. Many malware classification models, empowered by machine and/or deep learning, achieve superior accuracies for predicting malware types. Machine learning-based models need to do heavy feature engineering work, which affects the performance of the models greatly. On the other hand, deep learning-based models require less effort in feature engineering when compared to that of the machine learning-based models. However, traditional deep learning architectures components, such as max and average pooling, cause architecture to be more complex and the models to be more sensitive to data. The capsule network architectures, on the other hand, reduce the aforementioned complexities by eliminating the pooling components. Additionally, capsule network architectures based models are less sensitive to data, unlike the classical convolutional neural network architectures. This paper proposes an ensemble capsule network model based on the bootstrap aggregating technique. The proposed method is tested on two widely used, highly imbalanced datasets (Malimg and BIG2015), for which the-state-of-the-art results are well-known and can be used for comparison purposes. The proposed model achieves the highest F-Score, which is 0.9820, for the BIG2015 dataset and F-Score, which is 0.9661, for the Malimg dataset. Our model also reaches the-state-of-the-art, using 99.7% lower the number of trainable parameters than the best model in the literature.Conference Object Citation - WoS: 5Citation - Scopus: 9Website Category Classification Using Fine-Tuned Bert Language Model(Institute of Electrical and Electronics Engineers Inc., 2020) Demirkıran, Ferhat; Çayır, Aykut; Ünal, Uğur; Dağ, HasanThe contents on the Word Wide Web is expanding every second providing web users a rich content. However, this situation may cause web users harm rather than good due to its harmful or misleading information. The harmful contents can contain text, audio, video, or image that can be about violence, adult contents, or any other harmful information. Especially young people may readily be affected with these harmful information psychologically. To prevent youth from these harmful contents, various web filtering techniques, such as keyword filtering, Uniform Resource Locator (URL) based filtering, Intelligent analysis, and semantic analysis, are used. We propose an algorithm that can classify websites, which may contain adult contents, with 67.81% (BERT) accuracy among 32 unique categories. We also show that a BERT model gives higher accuracy than both the Sequential and Functional API models when used for text classification.
