Kocaeli University Research Information System
Not a member yet
80615 research outputs found
Sort by
Large Language Models for Machine Learning Design Assistance: Prompt-Driven Algorithm Selection and Optimization in Diverse Supervised Learning Tasks
No full textLarge language models (LLMs) are playing an increasingly important role in data science applications. In this study, the performance of LLMs in generating code and designing solutions for data science tasks is systematically evaluated based on different real-world tasks from the Kaggle platform. Models from different LLM families were tested under both default settings and configurations with hyperparameter tuning (HPT) applied. In addition, the effects of few-shot prompting (FSP) and Tree of Thought (ToT) strategies on code generation were compared. Alongside technical metrics such as accuracy, F1 score, Root Mean Squared Error (RMSE), execution time, and peak memory consumption, LLM outputs were also evaluated against Kaggle user-submitted solutions, leaderboard scores, and two established AutoML frameworks (auto-sklearn and AutoGluon). The findings suggest that, with effective prompting strategies and HPT, models can deliver competitive results on certain tasks. The ability of some LLMS to suggest appropriate algorithms reveals that LLMs can be seen not only as code generators, but also as systems capable of designing machine learning (ML) solutions. This study presents a comprehensive analysis of how strategic decisions such as prompting methods, tuning approaches, and algorithm selection, affect the design of LLM-based data science systems, offering insights for future hybrid human–LLM systems
Aronia melanocarpa yapraklarından kallus üretimi ve metanolik ekstraktının bazı bakteri ve funguslar üzerine antimikrobiyal aktivitesinin değerlendirilmesi
No full textAKILLI LİMANLAR VE DİJİTAL DÖNÜŞÜM: SÜRDÜRÜLEBİLİR LOJİSTİK SİSTEMLERİNE GEÇİŞ
No full textSürdürülebilirlik, limanların çevresel, sosyal ve ekonomik boyutlarını dengede tutarak ekolojik hedeflere ulaşmasını ve kalıcı etkinlik sağlamasını mümkün kılan bir kavramdır. Günümüz koşullarında, liman işletmeleri artan çevresel baskılar ve toplumsal talepler doğrultusunda yalnızca finansal performanslarını değil, aynı zamanda sürdürülebilirlik performanslarını da geliştirmek zorundadır. Bu bağlamda, nesnelerin interneti (IoT), büyük veri analitiği, yapay zekâ ve otomasyon sistemlerini kapsayan dijital dönüşüm kavramı enerji verimliliğini artırmak, israfı en aza indirmek ve liman operasyonlarında şeffaflığı güçlendirmek açısından büyük önem taşımaktadır. Dijitalleşme süreçleri, liman paydaşları arasındaki işbirliği dinamiklerini, veri alışverişi mekanizmalarını ve yönetişim çerçevelerini köklü biçimde dönüştürmektedir. Bu dönüşüm, sürdürülebilir tedarik zincirlerinin gelişimini kolaylaştırmakta ve yeşil lojistik uygulamalarının yaygınlaşması için sağlam bir zemin oluşturmaktadır. Dolayısıyla, dijital teknolojilerin sürdürülebilir liman yönetimiyle bütüncül biçimde entegrasyonu, çevresel bütünlüğün korunması ve ekonomik dayanıklılığın artırılması açısından önemli bir stratejik unsur olarak değerlendirilmektedir. Bu araştırma, sürdürülebilir liman operasyonları çerçevesinde dijital dönüşüm ve Endüstri 4.0 ile ilişkili teknolojilerin önemini titizlikle incelemekte ve akıllı liman sistemlerinde kullanılan yöntemleri detaylandırmaktadır. Limanların kalıcı çevresel sürdürülebilirliğini artırmak amacıyla ileri düzey bilgi teknolojilerinin ekolojik yönetim metodolojileriyle entegrasyonunu açıklamayı amaç edinmektedir. Çalışmada, teknolojik gelişmelerin operasyonel verimlilik, güvenlik, enerji tasarrufu ve liman operasyonlarındaki emisyonların azaltılması gibi kritik alanlarda önemli iyileştirmeler sağladığı gösterilmektedir. Ayrıca çalışma, akıllı liman teknolojilerinin kapsamlı şekilde uygulanmasının denizcilik endüstrisinde stratejik bir gereklilik olduğunu vurgulamakta; bu teknolojilerin entegrasyonunun ise yalnızca limanların rekabetçi konumunu güçlendirmekle kalmadığını, aynı zamanda çevresel dönüşüm hedeflerine ulaşılmasında da kritik bir rol oynadığını belirtilerek literatüre önemli bir katkı sunmaktadır.Sustainability is a concept that enables ports to achieve ecological goals and ensure lasting effectiveness by balancing their environmental, social, and economic dimensions. Under current conditions, port operators must improve not only their financial performance but also their sustainability performance in line with increasing environmental pressures and social demands. In this context, the concept of digital transformation, encompassing the Internet of Things (IoT), big data analytics, artificial intelligence, and automation systems, is of great importance in terms of increasing energy efficiency, minimizing waste, and strengthening transparency in port operations. This transformation facilitates the development of sustainable supply chains and creates a solid foundation for the widespread adoption of green logistics practices. Therefore, the comprehensive integration of digital technologies with sustainable port management is considered an important strategic element in terms of preserving environmental integrity and increasing economic resilience. This research meticulously examines the importance of digital transformation and Industry 4.0-related technologies within the framework of sustainable port operations and details the methods used in smart port systems. It aims to explain the integration of advanced information technologies with ecological management methodologies to enhance the long-term environmental sustainability of ports. The study demonstrates that technological developments provide significant improvements in critical areas such as operational efficiency, safety, energy savings, and reduction of emissions in port operations. Furthermore, the study emphasizes that the comprehensive implementation of smart port technologies is a strategic necessity in the maritime industry; it contributes significantly to the literature by stating that the integration of these technologies not only strengthens the competitive position of ports but also plays a critical role in achieving environmental transformation goals.</p
Sanatta Dijital Dönüşüm: Görsel Sanatlar Eğitiminde Teknoloji Entegrasyonuna Yönelik Eğilimler ve Bulgular
No full textComparison of Different Aperture Values in Dental Photography in Terms of Depth of Field, Sharpness, Diffraction, and Chromatic Aberration: A Preliminary Study
No full textEffect of H2S on oxidative steam reforming of biogas for syngas production over MgAl-supported Ni–Ce-based catalysts
No full textThis study investigates the effect of hydrogen sulfide (H2S) on the oxidative steam reforming (OSR) of biogas over particulate and monolithic NiCe/MgAl catalysts synthesized via the sol-gel method followed by the incipient wetness impregnation. Catalytic performance was evaluated in a flow reactor at 600 °C, 700 °C, and 800 °C under a constant space velocity of 45,000 mL gcat-1 h-1 and a CH4/CO2/O2/H2O molar feed ratio of 1/0.67/0.1/0.3, under H2S concentrations of 0, 12, and 50 ppm. Characterization was conducted using N2 physisorption, XRD, SEM, TGA, XPS and ICP-OES. Monolithic NiCe/MgAl (NCMA) exhibited reduced carbon deposition across all temperatures, whereas particulate NCMA achieved the highest CH4 (94%) and CO2 (80%) conversions at 800 °C. With 12 ppm H2S, particulate NCMA showed only a 10% decrease in CH4 conversion after 270 min. At 50 ppm H2S, both catalysts experienced significant deactivation, with CH4 conversion declining by approximately 50% after 270 min
Benchmarking Nanopore Sequencing for CLN2 (TPP1) Mutation Detection: Integrating Rapid Genomics and Orthogonal Validation for Precision Diagnostics
No full textCLN2 disease (neuronal ceroid lipofuscinosis type 2) is an ultra-rare lysosomal storage disorder caused by mutations in the TPP1/CLN2 gene, resulting in impaired tripeptidyl peptidase 1 (TPP1) activity. The timely initiation of enzyme replacement therapy is pivotal for attenuating progressive and irreversible neurodegeneration. This study aimed to benchmark the performance of Oxford Nanopore long-read sequencing (ONT-LRS) for targeted TPP1 mutation detection in a Turkish CLN2 cohort and to assess its concordance with orthogonal validation methods, including Sanger sequencing and enzymatic activity assays. Using a custom-designed primer panel, the entire TPP1 gene (6846 bp) was sequenced on the Oxford Nanopore (ONT) MinIon platform in seven clinically confirmed CLN2 index patients and sixteen unaffected family members. Detected variants were validated via Sanger sequencing and correlated with TPP1 enzyme activity in leucocytes and dried blood spots. Four pathogenic or likely pathogenic TPP1 variants were identified: c.622C>T (p.Arg208*), c.857A>G (p.Asn286Ser), c.1204G>T (p.Glu402*), and c.225A>G (p.Gln75=), along with fourteen additional benign variants. Variant allele frequencies were 50% for c.622C>T, 28.6% for c.1204G>T, 14.3% for c.857A>G, and 7.1% for c.225A>G. Notably, this is the first report to document the homozygous state of the c.857A>G variant and the compound heterozygous configuration of the c225A>G and c.622C>T variants in CLN2 patients, thereby expanding the known mutational landscape. In contrast, the globally common variant c.509-1G>C was not observed, suggesting regional variation in TPP1 mutation patterns. Consistent with the prior Turkish studies, c.622C>T (p.Arg208*) was the most prevalent variant, followed by c.1204G>T (p.Glu402*). TPP1 enzymatic activity was significantly reduced in all affected individuals (p < 0.0001), supporting the functional relevance of the identified variants. ONT-LRS offers a robust, cost-effective platform for high-resolution analysis of the TPP1 gene. Integrating molecular and biochemical data improves diagnostic precision and supports timely, targeted interventions for CLN2 disease, particularly in regions with high consanguinity and limited diagnostic infrastructure