Yıldız, Beytullah
Loading...
Name Variants
Yıldız, Beytullah
B.,Yildiz
Yildiz, B
B., Yildiz
B., Yıldız
Beytullah, Yildiz
Y.,Beytullah
Yildiz,B.
Y., Beytullah
Yıldız,B.
Beytullah, Yıldız
Yildiz, Beytullah
B.,Yıldız
B.,Yildiz
Yildiz, B
B., Yildiz
B., Yıldız
Beytullah, Yildiz
Y.,Beytullah
Yildiz,B.
Y., Beytullah
Yıldız,B.
Beytullah, Yıldız
Yildiz, Beytullah
B.,Yıldız
Job Title
Doçent Doktor
Email Address
beytullah.yildiz@atilim.edu.tr
ORCID ID
Scopus Author ID
Turkish CoHE Profile ID
Google Scholar ID
WoS Researcher ID
Scholarly Output
15
Articles
6
Citation Count
35
Supervised Theses
4
15 results
Scholarly Output Search Results
Now showing 1 - 10 of 15
Article Citation Count: 6Improving word embedding quality with innovative automated approaches to hyperparameters(Wiley, 2021) Yildiz, Beytullah; Yıldız, Beytullah; Tezgider, Murat; Yıldız, Beytullah; Software EngineeringDeep learning practices have a great impact in many areas. Big data and significant hardware developments are the main reasons behind deep learning success. Recent advances in deep learning have led to significant improvements in text analysis and classification. Progress in the quality of word representation is an important factor among these improvements. In this study, we aimed to develop word2vec word representation, also called embedding, by automatically optimizing hyperparameters. Minimum word count, vector size, window size, negative sample, and iteration number were used to improve word embedding. We introduce two approaches for setting hyperparameters that are faster than grid search and random search. Word embeddings were created using documents of approximately 300 million words. We measured the quality of word embedding using a deep learning classification model on documents of 10 different classes. It was observed that the optimization of the values of hyperparameters alone increased classification success by 9%. In addition, we demonstrate the benefits of our approaches by comparing the semantic and syntactic relations between word embedding using default and optimized hyperparameters.Conference Object Citation Count: 0Developing and Evaluating a Model-Based Metric for Legal Question Answering Systems(Institute of Electrical and Electronics Engineers Inc., 2023) Bakir,D.; Yildiz,B.; Aktas,M.S.; Software EngineeringIn the complicated world of legal law, Question Answering (QA) systems only work if they can give correct, situation-aware, and logically sound answers. Traditional evaluation methods, which rely on superficial similarity measures, can't catch the complex accuracy and reasoning needed in legal answers. This means that evaluation methods need to change completely. To fix the problems with current methods, this study presents a new model-based evaluation metric that is designed to work well with legal QA systems. We are looking into the basic ideas that are needed for this kind of metric, as well as the problems of putting it into practice in the real world, finding the right technological frameworks, creating good evaluation methods. We talk about a theory framework that is based on legal standards and computational linguistics. We also talk about how the metric was created and how it can be used in real life. Our results, which come from thorough tests, show that our suggested measure is better than existing ones. It is more reliable, accurate, and useful for judging legal quality assurance systems. © 2023 IEEE.Conference Object Citation Count: 0Reinforcement Learning for Intrusion Detection(Springer Science and Business Media Deutschland GmbH, 2023) Saad,A.M.S.E.; Yildiz,B.; Software EngineeringNetwork-based technologies such as cloud computing, web services, and Internet of Things systems are becoming widely used due to their flexibility and preeminence. On the other hand, the exponential proliferation of network-based technologies exacerbated network security concerns. Intrusion takes an important share in the security concerns surrounding network-based technologies. Developing a robust intrusion detection system is crucial to solving the intrusion problem and ensuring the secure delivery of network-based technologies and services. In this paper, we propose a novel approach using deep reinforcement learning to detect intrusions to make network applications more secure, reliable, and efficient. As for the reinforcement learning approach, Deep Q-learning is used alongside a custom-built Gym environment that mimics network attacks and guides the learning process. The NSL-KDD dataset is used to create the reinforcement learning environment to train and evaluate the proposed model. The experimental results show that our proposed reinforcement learning approach outperforms other related solutions in the literature, achieving an accuracy that exceeds 93%. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.Article Citation Count: 0Beyond ROUGE: A Comprehensive Evaluation Metric for Abstractive Summarization Leveraging Similarity, Entailment, and Acceptability(World Scientific Publ Co Pte Ltd, 2024) Briman, Mohammed Khalid Hilmi; Yıldız, Beytullah; Yildiz, Beytullah; Yıldız, Beytullah; Software EngineeringA vast amount of textual information on the internet has amplified the importance of text summarization models. Abstractive summarization generates original words and sentences that may not exist in the source document to be summarized. Such abstractive models may suffer from shortcomings such as linguistic acceptability and hallucinations. Recall-Oriented Understudy for Gisting Evaluation (ROUGE) is a metric commonly used to evaluate abstractive summarization models. However, due to its n-gram-based approach, it ignores several critical linguistic aspects. In this work, we propose Similarity, Entailment, and Acceptability Score (SEAScore), an automatic evaluation metric for evaluating abstractive text summarization models using the power of state-of-the-art pre-trained language models. SEAScore comprises three language models (LMs) that extract meaningful linguistic features from candidate and reference summaries and a weighted sum aggregator that computes an evaluation score. Experimental results show that our LM-based SEAScore metric correlates better with human judgment than standard evaluation metrics such as ROUGE-N and BERTScore.Conference Object Citation Count: 9Improving Text Classification with Transformer(Institute of Electrical and Electronics Engineers Inc., 2021) Soyalp,G.; Alar,A.; Ozkanli,K.; Yildiz,B.; Software EngineeringHuge amounts of text data are produced every day. Processing text data that accumulates and grows exponentially every day requires the use of appropriate automation tools. Text classification, a Natural Language Processing task, has the potential to provide automatic text data processing. Many new models have been proposed to achieve much better results in text classification. The transformer model has been introduced recently to provide superior performance in terms of accuracy and processing speed in deep learning. In this article, we propose an improved Transformer model for text classification. The dataset containing information about the books was collected from an online resource and used to train the models. We witnessed superior performance in our proposed Transformer model compared to previous state-of-art models such as L S T M and CNN. © 2021 IEEEArticle Citation Count: 14Text classification using improved bidirectional transformer(Wiley, 2022) Tezgider, Murat; Yıldız, Beytullah; Yildiz, Beytullah; Aydin, Galip; Yıldız, Beytullah; Software EngineeringText data have an important place in our daily life. A huge amount of text data is generated everyday. As a result, automation becomes necessary to handle these large text data. Recently, we are witnessing important developments with the adaptation of new approaches in text processing. Attention mechanisms and transformers are emerging as methods with significant potential for text processing. In this study, we introduced a bidirectional transformer (BiTransformer) constructed using two transformer encoder blocks that utilize bidirectional position encoding to take into account the forward and backward position information of text data. We also created models to evaluate the contribution of attention mechanisms to the classification process. Four models, including long short term memory, attention, transformer, and BiTransformer, were used to conduct experiments on a large Turkish text dataset consisting of 30 categories. The effect of using pretrained embedding on models was also investigated. Experimental results show that the classification models using transformer and attention give promising results compared with classical deep learning models. We observed that the BiTransformer we proposed showed superior performance in text classification.Conference Object Citation Count: 0A Novel Use of Reinforcement Learning for Elevated Click-Through Rate in Online Advertising(Ieee Computer Soc, 2023) Haider, Umair; Yildiz, Beytullah; Software EngineeringEfficiently predicting Click-through Rate (CTR) is crucial for the success of online advertising. Traditional methods often struggle to adapt to the dynamic nature of user preferences and the evolving relevance of advertisements. In this study, we propose a novel Reinforcement Learning (RL) approach for CTR prediction, leveraging OpenAI Gym and the Thompson Sampling algorithm. Our approach dynamically estimates CTR, cleverly adapting to the ever-changing landscape of user preferences and advertisement relevance. Results showcase the exceptional performance of Thompson Sampling in CTR prediction, sur-passing other RL methods with a remarkable 10% higher confidence level. This emphasizes the significant potential of our RL approach in optimizing the selection of online advertisements.Conference Object Citation Count: 1Enhancing Image Resolution with Generative Adversarial Networks(Institute of Electrical and Electronics Engineers Inc., 2022) Yildiz,B.; Software EngineeringSuper-resolution is the process of generating high-resolution images from low-resolution images. There are a variety of practical applications used in real-world problems such as high-definition content creation, surveillance imaging, gaming, and medical imaging. Super-resolution has been the subject of many researches over the past few decades, as improving image resolution offers many advantages. Going beyond the previously presented methods, Generative Adversarial Networks offers a very promising solution. In this work, we will use the Generative Adversarial Networks-based approach to obtain 4x resolution images that are perceptually better than previous solutions. Our extensive experiments, including perceptual comparison, Peak Signal-to-Noise Ratio, and classification success metrics, show that our approach is quite promising for image super-resolution. © 2022 IEEE.Article Citation Count: 0Daha İyi Dağıtımla İyileştirilmiş Dengesiz Veriler Üzerinde Derin Öğrenme ile Verimli Metin Sınıflandırması(2022) Yıldız, Beytullah; Yıldız, Beytullah; Software EngineeringTeknolojik gelişmeler ve internetin yaygınlaşması, günlük olarak üretilen verilerin katlanarak artmasına neden olmaktadır.\rBu veri tufanının önemli bir kısmı sosyal medya, iletişim araçları, müşteri hizmetleri gibi uygulamalardan gelen metin\rverilerinden kaynaklanmaktadır. Bu büyük miktarda metin verisinin işlenmesi otomasyona ihtiyaç duymaktadır. Son\rzamanlarda metin işlemede önemli başarılar elde edilmiştir. Özellikle derin öğrenme uygulamaları ile metin sınıflandırma\rperformansı oldukça tatmin edici hale gelmiştir. Bu çalışmada, metin sınıflandırma başarısını daha da artırmak için veri\rdengesizliği sorununu azaltan yenilikçi bir veri dağıtım algoritması önerdik. Deney sonuçları, veri dağılımını optimize eden\ralgoritma ile sınıflandırma doğruluğunda yaklaşık %3,5 ve F1 puanında 3'ün üzerinde bir iyileşme olduğunu göstermektedir.Article Citation Count: 1Reinforcement learning using fully connected, attention, and transformer models in knapsack problem solving(Wiley, 2022) Yildiz, Beytullah; Yıldız, Beytullah; Yıldız, Beytullah; Software EngineeringKnapsack is a combinatorial optimization problem that involves a variety of resource allocation challenges. It is defined as non-deterministic polynomial time (NP) hard and has a wide range of applications. Knapsack problem (KP) has been studied in applied mathematics and computer science for decades. Many algorithms that can be classified as exact or approximate solutions have been proposed. Under the category of exact solutions, algorithms such as branch-and-bound and dynamic programming and the approaches obtained by combining these algorithms can be classified. Due to the fact that exact solutions require a long processing time, many approximate methods have been introduced for knapsack solution. In this research, deep Q-learning using models containing fully connected layers, attention, and transformer as function estimators were used to provide the solution for KP. We observed that deep Q-networks, which continued their training by observing the reward signals provided by the knapsack environment we developed, optimized the total reward gained over time. The results showed that our approaches give near-optimum solutions and work about 40 times faster than an exact algorithm using dynamic programming.
