IYTE GCRIS Database (Izmir Institute of Technology)
Not a member yet
11968 research outputs found
Sort by
Community Detection for Large Graphs on GPUs With Unified Memory
No full textWhile GPUs accelerate applications from different domains with different characteristics, processing large datasets gets infeasible on target systems with limited device memory. Unified memory support makes it possible to work with data larger than available GPU memory. However, page migration overhead for executions with irregular memory access patterns, like graph processing workloads, induces severe performance degradation. While memory hints help to deal with page movements by keeping data in suitable memory spaces, coarse-grain configurations can still not avoid migrations for executions having diverse data structures. In this work, we target the state-of-the-art CUDA implementation of the Louvain community detection algorithm and evaluate the impacts of the fine-grained unified memory hints on the performance. Our experimental evaluation shows that memory hints configured for specific data structures reveal significant performance improvements and enable us to work efficiently with large graphs
Graphlet mining in big data
No full textThis thesis explores graphlet counting algorithms, which are crucial for understanding the structural principles of complex networks such as bioinformatics, social networks, and network model evaluation. Counting graphlets in large networks is computationally challenging due to the combinatorial explosion of possibilities, particularly for larger graphlet sizes. To address this, we focus on clique graphlets, fully connected subgraphs, which reveal critical patterns in areas like protein structure analysis, social network modeling, community detection, and spam detection. Counting k-cliques (subgraphs with nodes) becomes infeasible for large datasets and high values. Existing exact and approximate algorithms struggle with large , often failing when exceeds 10. To tackle these limitations, we propose BDAC (Boundary-Driven Approximations of K-Cliques), a novel algorithm that efficiently approximates k-clique counts using classical extremal graph theorems. BDAC uniquely provides lower and upper bounds for k-clique counts at both local (per vertex) and global levels, making it particularly suited for large, dense graphs with high values. Unlike existing methods, the algorithm's complexity remains unaffected by the value of . We validate BDAC's efficiency and scalability through extensive comparisons with leading algorithms on diverse datasets, spanning k values from minor (e.g., 8) to large (e.g., 50). Parallelization techniques enhance its performance, making it highly scalable for analyzing large and dense networks. BDAC offers a significant advancement in k-clique counting, enabling the analysis of previously considered computationally intractable networks.Bu tez, biyoenformatik, sosyal ağlar ve ağ modeli değerlendirmesi gibi karmaşık ağların yapısal prensiplerini anlamak için kritik öneme sahip olan alt çizge sayma algoritmalarını incelemektedir. Büyük ağlarda alt çizgelerin sayılması, özellikle daha büyük alt çizge boyutları için olasılıkların kombinatoryel patlaması nedeniyle hesaplama açısından zorludur. Bu zorlukları ele almak için, protein yapısı analizi, sosyal ağ modelleme, topluluk tespiti ve spam tespiti gibi alanlarda kritik desenleri ortaya çıkaran tam bağlı alt çizgeler olan k-klik alt çizgelerine odaklanıyoruz. K-klik'lerin ( düğümlü alt çizgeler) sayılması, büyük veri kümeleri ve yüksek değerleri için uygulanamaz hale gelmektedir. Mevcut kesin ve yaklaşık algoritmalar, 10'u aştığında genellikle başarısız olur. Bu sınırlamaların üstesinden gelmek için, klasik ekstremal çizge teoremlerini kullanarak k-klik sayılarını verimli bir şekilde yaklaşık olarak hesaplayan yenilikçi bir algoritma olan BDAC'ı (K-kliklerin Sınır Tabanlı Yaklaşımı) öneriyoruz. BDAC, k-klik sayımları için hem yerel (düğüm bazında) hem de küresel seviyelerde benzersiz bir şekilde alt ve üst sınırlar sağlayarak, özellikle yüksek değerlerine sahip büyük ve yoğun çizgeler için son derece uygundur. Mevcut yöntemlerin aksine, algoritmanın karmaşıklığı değerinden etkilenmez. BDAC'ın verimliliğini ve ölçeklenebilirliğini, küçük (ör. 8) ile büyük (ör. 50) arasında değişen değerlerini kapsayan çeşitli veri kümeleri üzerinde önde gelen algoritmalarla yapılan kapsamlı karşılaştırmalarla doğruluyoruz. Paralelleştirme teknikleri, performansını daha da artırarak, büyük ve yoğun çizgelerin analizinde oldukça ölçeklenebilir hale getirmektedir. BDAC, k-klik sayımı konusunda önemli bir ilerleme sunarak, daha önce hesaplama açısından ulaşılamaz kabul edilen çizgelerin analizine olanak tanımaktadır
Application of fourier transform infrared spectroscopy and molecular techniques in the in vitro study of iron deficiency anemia
No full textBu tez özelinde ele alınacak olan hastalık modeli olan demir eksikliği anemisinin (Iron Deficiency Anemia (IDA)) sebep olduğu fizyolojik değişimleri ve moleküler düzeydeki mekanizmaları anlamak, bu noktada öncü bilgiler elde etmek ve tedavi sonuçlarının tahmin edilmesi için uygun maliyetli ve pratik olan vibrasyonel spektroskopi teknikleri kullanılmıştır. İleriki çalışmalarda anemiye yönelik tasarlanan fonksiyonel gıda ; ilaç ürünlerinin anemi üzerindeki etkilerinin veya bu ürünlerin farklı parametlerinin hızlı, pratik ve uygun maliyetli bir şekilde analiz edilmesinde kullanılmak üzere, IDA üzerinde iyileştirici etkisi bilinen ve takviye olarak kullanımı yaygın olan demir formlarıyla tedavi edilen hücrelerden elde edilen spektroskopik veriler kullanılarak bir model oluşturulmuştur. Oluşturulan modelin ileriki çalışmalardaki kullanımını test etmek amacıyla geliştirilen demirin serbest formunu azaltan ve biyoyararlanımını artıran fonksiyonel bir gıda bileşeni olarak işlev görmesi amaçlanan protein-demir kompleksleri vibrasyonel spektroskopi teknikleri ile analiz edilmiştir. Elde edilen veriler oluşturulan modele entegre edilerek protein-demir kompleksinin terapötik etkisi vibrasyonel spektroskopi teknikleri kullanılarak gösterilmiştir. Vibrasyonel spektroskopi teknikleri kullanılarak gösterilen terapötik etkiler, moleküler ve genetik metotlar kullanılarak elde edilen sonuçlarla doğrulanmıştır.In this thesis, FTIR spectroscopy technique has been used to understand the physiological changes and molecular mechanisms caused by iron deficiency anemia (IDA) and to obtain leading information and predict treatment outcomes at an affordable cost. A model has been created using spectroscopic data obtained from cells treated with iron forms that are known to have a therapeutic effect on IDA and are commonly used as supplements. This model can be used to quickly, practically, and affordably analyze the effects of functional food and drug products designed for anemia or different parameters of these products. To test the use of the model in future studies, anemic cells are treated with protein-iron complexes that reduce the free form of iron and increase its bioavailability, as a functional food component, have been analyzed using FTIR spectroscopy technique. The therapeutic effect of the protein-iron complex has been demonstrated using FTIR spectroscopy technique and confirmed by results obtained using molecular and genetic methods
k-Clique counting on large scale-graphs: a survey
No full textClique counting is a crucial task in graph mining, as the count of cliques provides different insights across various domains, social and biological network analysis, community detection, recommendation systems, and fraud detection. Counting cliques is algorithmically challenging due to combinatorial explosion, especially for large datasets and larger clique sizes. There are comprehensive surveys and reviews on algorithms for counting subgraphs and triangles (three-clique), but there is a notable lack of reviews addressing k-clique counting algorithms for k > 3. This paper addresses this gap by reviewing clique counting algorithms designed to overcome this challenge. Also, a systematic analysis and comparison of exact and approximation techniques are provided by highlighting their advantages, disadvantages, and suitability for different contexts. It also presents a taxonomy of clique counting methodologies, covering approximate and exact methods and parallelization strategies. The paper aims to enhance understanding of this specific domain and guide future research of k-clique counting in large-scale graphs
Dynamic Compression of Metal Syntactic Foam-Filled Aluminum Tubes
No full textThe current research investigates the compressive properties of metal syntactic foam (MSF)-filled tubes at dynamic loads with an impact velocity of 4 m/s. For this purpose, A356 aluminum alloy syntactic foams were prepared using an infiltration casting technique with an incorporation of expanded perlite (EP) filler particles. The study involves the testing and comparison of both MSF samples and MSF-filled tubes under dynamic loading scenarios. In the case of MSF-filled tubes, aluminum tubes are either fully filled (FFT) or half-filled (HFT) with MSFs. The manufactured foams and foam cores have a similar macroscopic density across all tested samples. Under dynamic loading, the MSF, HFT, and FFT samples exhibit distinct and different deformation mechanisms. In MSFs, dynamic compression is controlled by shearing of the sample, whereas in HFTs and FFTs, dynamic deformation occurs through the folding and buckling of the tubes, accompanied by partial deformation of the MSF cores
Holistic Managements of Textile Wastewater Through Circular, Greener and Eco-Innovative Treatment Systems Developed by Minimal To Zero Liquid Discharge
No full textNew pragmatic and viable solutions to reduce or prevent discharge and to protect reserves are currently among the top-prioritised research for cleaner, circular, and resource-efficient use of industrial waters. So, the development of eco-sustainable water management is essential for green industrial development that will meet versatile and eco-sensitive regulatory standards, especially in water-intensive industries. Textile wastewater was reclaimed in semi to fully closed loops for minimal to zero liquid discharge. Concentrate-mixed wastewater was steadily treated in a hybrid membrane oxidation reactor at 60-80 % synergistic performances with remarkable UF fluxes of 96.4-820 L/m2h without any sludge discharge. Effluent was purified with 90-100 % removals and 20-80 L/m2h in nanofiltration and reverse osmosis. Due to Fenton-specific operation, more handling by ion exchange and neutralisation required to harvest membrane reuse waters and reactor discharge effluents with guaranteed Fe and pH. All-in-one system simulations indicated that high quality reuse waters are produced by 99.9 % efficiency and 98 and 100 % savings in iron and acid but 20-51 % more oxidant through concentrate recycling and regenerant reuse. It was also revealed that reactor effluents can be released to the sea or conventional biological treatment or can be eco-sustainably exploited for in-situ chemical and ex-situ bio-induced recovery of vivianite. This research demonstrates that how textile wastewater can be managed holistically by liquid discharge approaches from 50 % minimal to 99.9 % zero just in two-step, i.e. pretreatment and preconcentration, with consumable minimisation and valuable waste recovery through the eco-innovative systems which are developed as circular, greener, and sludge-free compatible with sustainable development goals
Assessment of the Validity and Reliability of Edinburgh Postpartum Depression Scale in Turkish Men
No full textObjectives: Perinatal depression (PD) affects individuals during pregnancy and early parenthood, resembling major depression. Recent research highlights paternal perinatal depression (PPD) in fathers. PPD has adverse effects on fathers and their children. This study assesses the Turkish version of the Edinburgh Postnatal Depression Scale (EPDS) for Turkish fathers, aiming to provide a tool for PPD identification. Methods: This methodological study validates the EPDS for Turkish fathers and explores associations with demographic and psychosocial factors. The study involved 295 fathers with infants aged 2 weeks to 12 months. The EPDS, originally designed for perinatal depression and validated in Turkish women, was used. Fathers completed a participant information questionnaire, the EPDS, and the Beck Depression Inventory (BDI) during clinic visits. Data on sociodemographic factors, paternal roles, and pregnancy and postpartum support were collected. Mothers also completed the EPDS. Descriptive statistics, exploratory factor analysis, confirmatory factor analysis, and correlation tests were used. Results: The study included fathers with an average age of 30.5 years, mostly with a high school education or higher. The EPDS had a mean total score of 3.1. Factor analysis suggested a three-factor structure for the EPDS in Turkish fathers, including anhedonia, anxiety, and depression. Confirmatory factor analysis validated the three-factor structure, with acceptable model fit indices. Positive correlations were found between fathers' EPDS scores, maternal EPDS scores, and paternal BDI scores. The EPDS effectively discriminated between different levels of depression severity. Various factors, such as education level and lack of support during pregnancy and after childbirth, were associated with higher EPDS scores. Conclusions: These findings emphasize the significance of assessing and addressing PPD in fathers, supporting the use of the EPDS as a valid tool in the Turkish context. The three-factor structure aligns with international research, highlighting the importance of a multi-dimensional approach to PPD assessment. Early intervention can mitigate PPD's impact on fathers, mothers, and children, benefiting mental health and well-being. © 2024 Walter de Gruyter GmbH, Berlin/Boston
Computing a Parametric Reveals Relation for Bounded Equal-Conflict Petri Nets
No full textIn a distributed system, in which an action can be either “hidden” or “observable”, an unwanted information flow might arise when occurrences of observable actions give information about occurrences of hidden actions. A collection of relations, i.e. reveals and its variants, is used to model such information flow among transitions of a Petri net. This paper recalls the reveals relations defined in [3], and proposes an algorithm to compute them on bounded equal-conflict PT systems, using a smaller structure than the one defined in [3]. © 2024, The Author(s), under exclusive license to Springer-Verlag GmbH, DE, part of Springer Nature
Doping Effect on the Anode Material Capability of 2d Bn Nanosheets
No full textIn this thesis, the potential of BNN surfaces doped with Al, Cl, Co, Fe, Ga, O, P, and S atoms as anode materials in K, Li, Mg, and Na ion batteries was investigated. Semi-empirical tight-binding combined with meta-dynamics methods and density functional theory were utilized to discover these properties. The effects of doping atoms on the electronic structure and geometry of BNN surfaces were also studied. Changes in the electronic structure and conductivity were reported by examining the HOMO-LUMO orbitals and the energy differences between these orbitals. Using previously reported experimental data and examining similar studies from the literature, the atoms to be doped were chosen. While vacancies at the sites of boron atoms in single-layer boron-nitride nanosheets were observed, vacancies formed by nitrogen atoms were not observed, indicating that boron vacancies are much more likely for the doping position. So that doping was performed on the boron atom. The level of quantum calculations used in this work was validated using experimental data. B3LYP/def2-SVP/D4/gCP level of theory is used for all calculations for BNN-nanosheets studied in this thesis. The bond lengths and the HOMO-LUMO energy difference were found to be nearly the same as the experimental data. The conductivity of the BNN surface was increased with the doping process. However, significant improvements are followed by doping of cobalt, iron, and sulfur atoms with 35%, 34%, and 26% alteration, respectively. For a suitable battery manufacture, the potential anode material should offer structures with high theoretical specific capacity, low anode electrode voltage, and minimal volume change between charged/discharged states. It was observed that none of the doped-BNN surfaces involved in this study were suitable for the use of anode material in magnesium ion batteries. On the other hand, they can be used as a negative electrode for potassium, lithium, and sodium batteries. Their capacity in lithium is better than Na and K batteries. Our results suggest that most of the doped BNN surface with ions studied in this thesis could be used as anode materials. However, none of them owns a better battery capacity than classic lithium batteries.Bu tezde, Al, Cl, Co, Fe, Ga, O, P, S atomları ile katkılandırılmış (doped) edilmiş BNN yüzeylerinin K, Li, Mg, Na iyon bataryalarında anot materyal olarak kullanılma potansiyeli incelenmiştir. Bu özellikleri keşfetmek için yarı deneysel sıkı bağlama yöntemleri ile meta-dinamik yöntemler ve yoğunluk fonksiyonel teorisi kullanılmıştır. Katkılanan atomlarının BNN yüzeylerinin elektronik yapısı ve geometrisi üzerindeki etkileri de incelenmiştir. Elektronik yapıdaki ve iletkenlikteki değişimleri, HOMO-LUMO orbitalleri ve bu orbitaller arasındaki enerji farklılıkları inceleyerek raporlanmıştır. Daha önce raporlanmış deneysel veriler kullanılarak ve literatürde benzer çalışmalar incelenerek katkılanacak atomlar seçilmiştir. Tek katmanlı bor-nitrit nano-tabakalarda boron atomlarının olduğu noktalarda boşluklar gözlemlenebilirken, nitrojen atomlarının oluşturduğu boşluklar gözlemlenmemiştir. Bu durum, doping pozisyonu için boron boşluklarının çok daha olası olduğunu göstermektedir. Bu nedenle doping, boron atomu üzerinde gerçekleştirilmiştir. Bu çalışmada kullanılan kuantum hesaplamaların seviyesi deneysel veriler kullanılarak doğrulanmıştır. BNN-nano-tabakalar için tüm hesaplamalarda B3LYP/def2-SVP/D4/gCP teori seviyesi kullanılmıştır. Bağ uzunlukları ve HOMO-LUMO enerji farkı, deneysel verilerle neredeyse aynı bulunmuştur. BNN yüzeyinin iletkenliği doping işlemi ile artırılmıştır. Ancak, kobalt, demir ve sülfür atomlarıyla katkılama sırasıyla %35, %34 ve %26'lık önemli iyileştirmeler sağlamıştır. Uygun bir pil üretimi için, anot materyalin yüksek teorik spesifik kapasiteye, düşük anot elektrot voltajına ve yüklü/yüksüz durumlar arasında minimal hacim değişimine sahip olmalıdır. Hiçbir katkılı-BNN yüzeyinin magnezyum iyon pillerinde kullanılmak için uygun olmadığı gözlemlenmiştir. Öte yandan, potasyum pilleri, lityum ve sodyum pillerine benzer performans göstermiş olmasına rağmen lityum piller en iyi performansı sergilemiştir. Sonuçlarımız, bu tezde incelenen iyonlarla katkılanmış çoğu BNN yüzeyinin anot materyali olarak kullanılabileceğini göstermektedir. Ancak, hiçbiri klasik lityum pillerinden daha iyi bir pil kapasitesine sahip değildir
Ankos Yazışma Kuralları Üye - Vts - Yayıncı İlişkileri
No full text02-05 Mart 2024 tarihleri arasında Uluslararası Final Üniversitesi ev sahipliğinde Kıbrıs'ta gerçekleşen ANKOS 18. Gönüllü Çalıştayı kapsamında sunulmak üzere hazırlanan çalışmadır