LEE- Bilgisayar Mühendisliği-Doktora
Bu koleksiyon için kalıcı URI
Gözat
Son Başvurular
1 - 5 / 43
-
ÖgeEnergy efficient resource management in cloud datacenters(Graduate School, 2023-07-11)We propose an energy efficient resource allocation approach that integrates Holt Winters forecasting model for optimizing energy consumption while considering performance in a cloud computing environment. The approach is based on adaptive decision mechanism for turning on/off machines and detection of over utilization. By this way, it avoids performance degradation and improves energy efficiency. The proposed model consists of three functional modules, a forecasting module, a workload placement module along with physical and virtual machines, and a monitoring module. The forecasting module determines the required number of processing unit (Nr) according to the user demand. It evaluates the number of submitted workloads (NoSW), mean execution time of submitted workloads in interval and mean CPU requirements of them to calculate approximately total processing requirement (APRtotal). These three values are forecasted separately via forecasting methodologies namely Holt Winters (HW) and Auto Regressive Integrated Moving Average (ARIMA). The Holt Winters gives significantly better result in term of Mean Absolute Percentage Error (MAPE), since the time series include seasonality and trend. In addition, the interval is short and the long period to be forecasted, the ARIMA is not the right choice. The future demand of processing unit is calculated using these data. Therefore, the forecasting module is based on Holt Winters forecasting methodology with 8.85 error rate. Therefore, the forecasting module is based on the Holt Winters. Workload placement module is responsible for allocation of workloads to suitable VMs and allocation of these VMs to suitable servers. According to the information received from forecasting module, decision about turning a server on and off and placement for incoming workload is making in this module. The monitoring module is responsible for observing system status for 5 min. The consolidation algorithm is based on single threshold whether to decide that the server is over utilized. In other words, if the utilization ratio of CPU exceeds the predefined threshold, it means that the server is over utilized otherwise, the server is under load. If the utilization of server equals the threshold, the server is running at optimal utilization rate. Unlike other studies, overloading detection does not trigger VM migration. Overloading is undesirable since it causes performance degradation but, it can be acceptable under some conditions. To decide allocation of incoming workloads, this threshold is not only and enough parameter. Beside the threshold, the future demands are also considered as important as systems current state. The proposed algorithm also uses different parameters as remaining execution time of a workload, active number of servers (Na), required number of servers (Nr) besides efficient utilization threshold. The system can be instable with two cases; (1) Na is greater than Nr that means there are underutilized servers and it causes energy inefficiency (2) Nr is greater than Na, if new servers are not switched on, it causes over utilized servers and performance degradation. The algorithm is implemented and evaluated in CloudSim which is commonly preferred in the literature since, it provides a fair comparison between the proposed algorithm with previous approaches and it is easy to adapt and implementation. However, workloads come to the system in a static manner and the usage rates of the works vary depending on time. Our algorithms provide dynamically submission. Therefore, to make fair comparison, the benchmark code is modified to meet dynamic requirement by working Google Cluster Data via MongoDB integration. The forecasting module is based on Holt Winters as described before. Therefore, the approach is named Look-ahead Energy Efficient Allocation – Holt Winters (LAA-HW). If we knew the actual values instead of forecasted values, the system would give the result as Look-ahead Energy Efficient Allocation –Optimal (LAA-O). The proposed model uses Na and Nr parameters to decide the system's trend whether the system has active servers than required. If Na is greater than the Nr, incoming workloads are allocated on already active servers. It causes bottleneck for workloads with short execution time and less CPU requirement as the Google Tracelog workloads. The mean cpu requirement of a day and the mean execution time of a day are 3% and 1,13 min 32 respectively. It gives the small Nr value and it causes less number of received workload than Local Regression-Minimum Migration Time (LRMMT). The number of migration is zero in our approach. The energy consumption for switching on and off in our model is less in comparison with the migration model.
-
ÖgeNovel centrality, topology and hierarchical-aware link prediction in dynamic networks(Graduate School, 2023-09-05)The increasing availability of social network data has given rise to research devoted to solving problems associated with social network-related applications. However, the hugeness and complexity of relationships among social network elements render the prediction of links between the entities a challenging task. The previous research often focuses primarily on investigating local node connectivity data while ignoring other important network-characterizing properties. The key network-characterizing properties that are often underrated include network topology, node structural centrality roles, and network hierarchical information. Furthermore, whereas many real-world graphs change over time, several works assume static networks. In order to overcome these challenges, first, we compute several topological similarity-based convolution feature matrices by using various topological similarity metrics such as Common Neighbour, Jaccard Index, Adamic Adder, Salton Index, Resource Allocation, and Sørensen Index. We then utilize the resulting topological feature matrices to capture the prevailing topological information in the input graphs efficiently. Second, we leverage the strength centrality, a stronger variant of node degree, to conserve the node's centrality and the structural connectivity information in the network. In addition, we systematically aggregate such diverse features to yield quality higher-level feature representations. Lastly, we leverage an LSTM layer to capture the prevailing temporal information in the graph sequences. To learn the low dimensional node representations, first, we deployed a fully connected variational autoencoder that efficiently explores variations in the input graphs to learn high-quality node embeddings. Furthermore, we imposed centrality and topological constraints on the learning model to further enforce the preservation of the centrality and topological ınformation of input graphs in the learned embeddings. However, variational autoencoders have large computational time and memory requirements due large number of parameters characterizing the fully connected encoders and decoders, especially when they are applied on large networks. In order to extend our implementations to large datasets while minimizing the computational time and memory requirements, we adopted a Graph Convolution Network (GCN)-based implementation. The proposed Structural and Topological based geometric deep learning approach was evaluated on five real-world temporal social networks. Based on experimental results, on average, they yield a 4\% link prediction AUC improvement in link prediction accuracy, a small increment in training for each epoch (0.2s (10\%)), and a 56\% MSE reduction in centrality prediction when compared to the best benchmarks. The proposed end-to-end centrality and topological guided link prediction framework for dynamic networks preserve not only the centrality node roles and the topological information in the learned embeddings but also captures the prevailing temporal information in the dynamic networks. The models utilize node centrality and topological features to capture and conserve the network topology and the structural roles of nodes during embedding learning. Thus, obtaining pretty-quality embeddings that enhance the link prediction and centrality prediction accuracies. For all our proposed methods, we assess the impact of the various modules of the proposed models by comparing them with their variants that lack such modules, and we present and explain the results accordingly. In other related work, we introduce a Hierarchical and Centrality aware Polypharmacy Side Effect Prediction (HC-POSE) Model. We model side effect prediction as a link prediction task problem and leverage core decomposition to explore the prevailing hierarchical information in the heterogeneous protein-protein, protein-drug, and drug-drug interaction datasets. Following k-core decomposition, for each k-core subgraph produced, a node strength matrix is computed to store the centrality information of each subgraph. Then we systematically aggregate the obtained centrality with the k-core adjacency matrix to have higher-level diverse feature representations. We deployed a GCN-based auto-encoder to learn low-dimensional representations for the homogeneous sub-graphs and an RCGN-based auto-encoder for the heterogeneous subgraphs. Based on the experimental results, HC-POSE exhibited a 3\% accuracy improvement in POSE prediction as compared to the best baseline.
-
ÖgeRl based network deployment and resource management solutions for opportunistic wireless access for aerial networks in disaster areas and smart city applications(Graduate School, 2023-08-09)The growth in the mobile communication area changed the data traffic profile. In addition, the requirement for the deployed infrastructure has significantly changed. The available bandwidth and IP transformation in the mobile backend increased the peak traffic requirements, while the mobile nature of the users changed the required infrastructure over time. The commercial availability of unmanned aerial vehicles potentially addresses requirements changes within the infrastructure. However, its three-dimensional nature and operation range limitations due to limited battery introduce new problems. Topology control is a significant problem for unmanned aerial vehicle networks. The optimization of the network size for coverage is identical to the minimum set-cover problem. The minimum set-cover problem is NP-hard, even without the service-level agreements enforced within the communication networks. The solution sets provided for tailor-made applications prevent the scalability of aerial networks. The tailor-made solutions require the exact development cost for each new application target. Reinforcement learning provides an ideal solution for addressing requirements for multiple applications with a single development effort. The integration cost depends on data availability for training in reinforcement learning-based deployments. To this end, reinforcement learning is integrated into a central software-defined networking-based control entity to demonstrate the deployment cycle of the aerial network. In addition, the solution's effectiveness is proved by comparing the quality of service, coverage, and power consumption results with existing literature. Furthermore, the application area of the reinforcement learning is extended to wireless channel selection to address the physical resource assignment problem. The development cost of the model has been the availability of the data. The integration of the new application is demonstrated in the simulation tool to measure the cost. In addition, smart-city application for the aerial network in distributed architecture is simulated with this implementation. Overall, this thesis conducts a survey of the existing literature on the challenges of aerial networks. In addition, the reinforcement learning integration tool is developed in a simulation format. Finally, the disaster area and smart-city applications are implemented to measure the applicability of the hypothesis. The comparison results revealed that reinforcement learning-based aerial network topology control provides scalable performance for power consumption while satisfying the quality of service and coverage requirements of the network. In addition, the improvements in the physical resource allocation for opportunistic access on the wireless medium is proved in wireless channel selection deployment for the smart-city application.
-
ÖgeFog computing-based real-time emotion recognition using physiological signals(Graduate School, 2025-02-03)Emotion recognition plays a pivotal role in affective computing and human-computer interaction, with physiological signals such as ElectroEncephaloGram (EEG), ElectroCardioGram (ECG), ElectroMyoGram (EMG), respiration, and Galvanic Skin Response (GSR) are more reliable indicators of emotions. Unlike facial expressions, gestures, or speech signals, these physiological signal measures offer greater consistency in detecting emotions. This reliability arises from their reduced susceptibility to subjectivity and external factors, such as environmental noise and language barriers. There are numerous uses for emotion recognition to monitor multi-user emotional states simultaneously in real-time such as smart homes, workplaces, education, healthcare, and entertainment. For smart homes, emotion recognition may be integrated to provide automation for family members by modifying lighting, music, and other environmental elements according to the overall emotional states of the households. For workplace wellness, employers may have obtained the ability to track the health of many employees, recognizing signs of stress and tiredness; thus, they can take timely actions for a healthier work environment. In educational settings, emotion recognition can provide educators with insights into student engagement and understanding for groups of students. That helps educators adjust their teaching strategies to suit the demands of each class and improves the quality of the learning process overall. In a clinical setting, emotion recognition can help healthcare professionals to monitor the emotional states of multiple patients simultaneously, receiving immediate alerts and providing timely interventions. In the entertainment industry, especially in multi-player gaming and virtual reality environments, emotion recognition may enhance the experience by adapting environmental settings to the emotional states of individual users in real time while ensuring low latency and high performance. This study explores the implementation of fog computing for real-time emotion recognition using physiological signals. Implementing an emotion recognition system in the Internet of Things (IoT) requires powerful computational resources. Therefore, existing studies highlight the potential of cloud and edge computing architectures for emotion recognition but reveal a gap in leveraging fog computing for scalable, real-time applications. Building on these insights, this study integrates fog computing to improve latency, response time, and scalability efficiency. Hence, sensor-to-cloud architectures face challenges like latency, high bandwidth requirements, and security concerns. Fog computing provides low latency, enhanced security, scalability, and efficient resource utilization by processing data closer to its source, ensuring reliable real-time performance for multi-user emotion recognition systems. The research adopts a comprehensive methodology, starting with the use of the DREAMER dataset, which contains EEG and ECG signals recorded from 23 participants under various emotional stimuli. Signals in the dataset were segmented into 3-second windows with a 2-second overlap. Then, the data in the windows were preprocessed, and 36 time-based statistical features were extracted from the signals. By merging the features obtained from 2-channel ECG and 14-channel EEG signals, a data vector of 576 features was obtained for each sample. The dataset was divided into training and testing sets to train and evaluate the machine learning models. Eight machine learning models are employed to predict emotions. Based on accuracy, recall, precision, and F1 score metrics, the Light Gradient Boosting Machine (LGBM) model demonstrated the best performance. Since the models are designed for real-time use, the inference time for a single sample was measured. The LGBM model provided the highest accuracy with an acceptable prediction time, making it the preferred choice for the proposed real-time system. The LGBM model's inference times on the worker and cloud devices were 7.26 ms and 2.85 ms with an accuracy of 85.27 %. Accuracy variations were analyzed as the number of features changed, and accuracy plateaued at around 85 % after 44 features were used. The maximum accuracy of 86.25 % was achieved using 136 features, resulting in an average response time of 61.01 ms. However, considering resource utilization and time performance requirements for real-time systems, the system was configured to use 48 features, yielding 84.85 % accuracy with a 33 % reduction in processing time. In the proposed system, the emotion recognition process runs every second and takes about 40 ms. This pre-trained machine learning model, based on 48 physiological signal features, was integrated into the fog computing architecture, allowing for real-time emotion recognition. A fog computing architecture is designed, comprising cloud, broker, and worker nodes, to manage data processing tasks efficiently. All computation unit components of the architecture were tested in real-time scenarios using the pre-trained model. Unit performances were evaluated based on metrics such as latency, queuing delay, jitter, total response time, and resource usage, with experimental results showing that a worker node can efficiently handle computational tasks. Overall, the emotion recognition procedure begins every second and takes approximately 40 ms, including 4 ms of latency and 33 ms of execution time. Therefore, the results demonstrate fog computing's superiority over edge and cloud computing. The proposed fog computing system also outperforms existing studies in response time for real-time feature extraction from physiological signals, confirming that the fog architecture is well-suited for a real-time emotion recognition system. The system's scalability and usability in a multi-user environment were also assessed. Cloud, broker, and worker devices supported up to 11, 5, and 6 users, respectively. Although the single worker device could serve fewer users than the cloud, the fog architecture addressed this by issue incorporating multiple, cost-effective worker devices. Furthermore, using multiple processes for data processing on the worker device enhanced multi-user capacity with an optimal configuration of 6 processes allowing a worker device to serve up to 11 users. The fog architecture, utilizing multiple workers, was also evaluated. Stress tests revealed that the system can scale to accommodate 30 users with 3 worker devices. The study's findings highlight the effectiveness of fog computing in real-time emotion recognition. There are three key results including that the LGBM model achieved the highest accuracy of 85.27 % and mean single inference time of 7.26 ms for worker device, outperforming other machine learning models. The second key result is that fog computing significantly reduced latency and response time compared to cloud-based architectures, ensuring faster processing of physiological signals. Lastly, the system with 3 worker nodes and a 6-process configuration demonstrated scalability, handling up to 30 users with stable response times of approximately 40 ms. Resource utilization was optimized, with fog nodes distributing computational workloads effectively to avoid bottlenecks. Several research areas remain unexplored in this study such as optimizing computational resources, expanding the dataset and emotion classes, conducting real-world experiments to assess the practical usability of the proposed system, and implementing deep learning models within the fog computing framework using larger datasets.
-
ÖgeUçtan uca derin öğrenme yaklaşımlarıyla Türkçe eşgönderge çözümlemesi(Lisansüstü Eğitim Enstitüsü, 2025-02-03)Eşgönderge Çözümlemesi (EÇ), bir doküman içinde yer alan, aynı gerçek dünya varlığının (ör. bir kişi, yer veya olay) temsili olan sözcükler (ifade) arasındaki göndergesel ilişkinin çözümlenmesidir. Doğal Dil İşleme (DDİ) alanının anlamsal katmanında önemli bir görev olarak yer alan EÇ, metnin bağlamını derinlemesine çözümleyerek, dokümanın doğru bir şekilde anlaşılmasına ve istenen bilgilerin doğru bir şekilde çıkarılmasına yardımcı olmaktadır. Bu görevde, aralarında ilişki çözümlemesi yapılacak sözcük veya sözcük öbekleri bir ifade olarak tanımlanır. Uçtan uca bir EÇ sistemi, iki aşamadan oluşur: 1) İfade Saptama, 2) İlişki Çözümleme. İfade saptama aşamasında, dokümandaki tüm göndergesel ifadeler tespit edilir. Sonrasında, bu ifadeler arasındaki ilişkiler çözümlenerek aynı gerçek dünya varlığını temsil eden ifadeler aynı ifade kümesi altında birleştirilir. Türkçe, biçim bilimsel açıdan oldukça zengin ve zamir düşürme özelliğine sahip bir dildir. Bu özellikleri, Türkçe metinlerde bazı zamirlerin metin içerisinde açıkça yer almamasına olanak tanımaktadır. Dolayısıyla, Türkçe için geliştirilen kapsamlı bir EÇ sisteminin, düşürülen zamirleri de birer ifade olarak ele alıp bu zamirlerin ilişki çözümlemesini yapması, Türkçe yazılmış bir metnin anlam bütünlüğünün doğru anlaşılabilmesi için son derece önemlidir. Düşen zamirlere ilişkin bilgiler, cümledeki başka bir sözcüğün biçim bilimsel katmanında yer almaktadır. Bu durum, sözcüklerin yalnızca orijinal formlarının değil, aynı zamanda biçim birim düzeyinde de incelenmesini zorunlu kılmaktadır; dolayısıyla, Türkçe EÇ problemi diğer dillere kıyasla daha karmaşık bir hale gelmektedir. EÇ literatüründe yer alan çalışmalar incelendiğinde, çalışmaların çoğunun İngilizce üzerinde gerçekleştirildiği görülmektedir. Dil bilimsel açıdan Türkçeye benzeyen diller üzerinde yapılan EÇ çalışmaların ise son yıllarda başladığı görülmektedir. Yukarıda belirtilen Türkçenin dil bilimsel yapısından kaynaklanan biçim birim düzeyinde eşgönderge çözümlemesi gerekliliği, İngilizce için geliştirilmiş sistemlerin Türkçe için doğrudan uygulanmasına olanak tanımamaktadır. Bu tez çalışmasının hedefi, Türkçenin dil bilimsel özelliklerini göz önünde bulunduran ve yapay sinir ağları yöntemlerinden faydalanan, uçtan uca ilk Türkçe EÇ modelini gerçekleştirmektir. Bu doğrultuda: 1) Türkçenin yapısı, düşürülen zamirler açısından incelenmiş ve bu bilgiler için EÇ görevine özgü bir etiketleme şeması önerilmiş ve düşürülmüş zamirlerin bu şema ile göndergesel ifadeler olarak etiketlendiği güncel bir Türkçe EÇ veri kümesi sunulmuş, 2) Derin öğrenme yöntemlerinden faydalanan, farklı EÇ yaklaşımları ile geliştirilmiş Türkçe EÇ modelleri geliştirilerek, modellerin başarımları karşılaştırılmış, 3) Önerilen Türkçe EÇ veri kümesinin, çok dilli EÇ çalışmalarında kullanılabilmesi için ilgili veri kümesi koleksiyonlarında yer almasına yönelik çalışmalar tamamlanmış, 4) Türkçeyi de kapsamına alan çok dilli EÇ modelleri geliştirilerek, modellerin başarımları karşılaştırılmış, 5) Sonuç olarak, kod çözücü mimarisine sahip büyük dil modellerinden faydalanan, talimatlı tabanlı eğitilen, çok dilli EÇ modellerinin Türkçe EÇ üzerinde en iyi performansı gösterdiği ortaya konmuştur. Ek olarak, çok dilli modeller üzerinde yapılan iyileştirmeler ile özellikle dil bilimsel açından Türkçeye benzeren başka dillerdeki EÇ performanslarında da artışlar gözlemlenmiştir. Tez çalışmasında, mevcut etiketli Türkçe EÇ veri kümesi iyileştirilmiş ve düşürülmüş zamirlerin göndergesel ilişkileri etiketlenerek literatürdeki en güncel Türkçe EÇ veri kümesi oluşturulmuştur. Türkçenin EÇ başarımına, farklı eşgönderge çözümlemesi yaklaşımlarıyla (ifade çifti, ifade sıralama, uçtan uca) geliştirilen yapay sinir ağları tabanlı modellerin etkisi incelenmiştir. Veri kümesinin kalitesi ve düşürülmüş zamir etiketlemelerinin Türkçe EÇ modellerinin başarısına etkisi araştırılmıştır. Ayrıca, derin öğrenme yöntemleriyle geliştirilen Türkçe EÇ modellerinde çizge sinir ağları katmanlarının kullanımı ve bunun performansa etkisi de incelenmiştir. Türkçe üzerinde eğitilen tek dilli modeller, çok dilli olarak genişletilerek diller arası transferin Türkçe EÇ başarımına etkisi değerlendirilmiştir. Bu aşamada, Türkçe ve diğer dillerdeki EÇ başarımlarının, dillerin birbirlerinden öğrendikleri bilgilerle nasıl etkilendiği incelenmiştir. Türkçenin biçim bilimsel zenginliği nedeniyle, dil bilimsel bilgilerin EÇ modellerinde öznitelik olarak kullanılmasının etkisi, Türkçe ve benzer dillerdeki çok dilli EÇ veri kümesi üzerinde araştırılmıştır. Son olarak, kod çözücü mimarisi ve talimat tabanlı yöntemle geliştirilen çok dilli EÇ modelinin Türkçe ve diğer dillerdeki başarımları incelenmiştir. Sonuçlar, derin öğrenme yöntemlerinin Türkçe EÇ başarımını artırdığını göstermektedir. Kaliteli verilerle eğitilen Türkçe EÇ modelleri daha iyi sonuçlar elde etmiştir. Ayrıca, düşürülmüş zamirlerin etiketlenmesi ve bu ifadeler üzerinde eğitim yapılması, genel EÇ başarımını olumlu etkilemiştir. Çizge sinir ağlarının Türkçe EÇ performansını iyileştireceği hipotezi doğrulanamamıştır. Çok dilli modeller geliştirerek, diller arası transferin Türkçe EÇ başarımına olan olumlu etkileri gösterilmiştir. Türkçe ve benzer dil bilimsel özelliklere sahip dillerin EÇ performanslarında, açıkça belirtilen biçimsel özniteliklerin kullanılmasının olumlu etkisi gözlemlenmiştir. Son olarak, talimat tabanlı eğitimle geliştirilen çok dilli Türkçe EÇ modeli ile büyük dil modellerinin gücünden faydalanarak hem Türkçe hem de çok dilli EÇ performanslarında iyileşme sağlanmıştır.