Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Atınç Yılmaz has a total of 37 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Individual Innovativeness, General Attitudes Toward Artificial Intelligence, and Perceptions of an AI-Equipped Robotic Nursing Support System Among Intensive Care Nurses: An Analytical Cross-Sectional Study

    by Hamiyet Kızıl, Atınç Yılmaz, Osman Çaylı

    Objective: To examine individual innovativeness and general attitudes toward artificial intelligence (AI), together with perceptions of a specific AI-equipped robotic nursing support system, among intensive care nurses with direct workplace exposure to the system. Methods: This single-center analytical cross-sectional study was conducted in Istanbul, Türkiye, using a census-oriented recruitment approach. Of the 120 intensive care nurses employed in participating intensive care units where the robotic nursing support system was available, 100 participated in the study (participation rate: 83.3%). Individual innovativeness was assessed with the Individual Innovativeness Scale (IIS), and general AI attitudes with the Positive and reverse-scored Negative subscales of the General Attitudes towards Artificial Intelligence Scale (GAAIS). Primary associations were examined using Pearson correlations with 95% confidence intervals (CIs). Robust multivariable models, rank-based sensitivity analyses, and false-discovery-rate-adjusted subgroup comparisons were added during peer review as supplementary, non-prespecified analyses; subgroup findings are exploratory. The study had no pre-exposure baseline or unexposed comparison group. Results: Mean IIS, Positive GAAIS, and reverse-scored Negative GAAIS scores were 71.59 ± 8.96, 48.56 ± 6.53, and 31.82 ± 4.57, respectively. Internal consistency was high for IIS (α = 0.905), Positive GAAIS (α = 0.886), and reverse-scored Negative GAAIS (α = 0.842). IIS was moderately associated with Positive GAAIS (r = 0.441, 95% CI 0.267–0.586; p < 0.001) and reverse-scored Negative GAAIS (r = 0.426, 95% CI 0.251–0.574; p < 0.001). In supplementary HC3-adjusted models, IIS remained associated with Positive GAAIS (B = 0.188, 95% CI 0.045–0.332; standardized β = 0.258; p = 0.011) and reverse-scored Negative GAAIS (B = 0.151, 95% CI 0.054–0.248; β = 0.297; p = 0.003). Previous AI/robot nursing education was also independently associated with both attitude dimensions. Perceived professional benefit (86%) and perceived workload reduction (86%) reflected nurses’ perceptions of AI/robots in nursing rather than objectively measured clinical outcomes. Additionally, 70% of participants reported satisfaction with the specific robotic nursing support system. Conclusions: Among robot-exposed intensive care nurses at one center, greater individual innovativeness was associated with more favorable general AI attitudes in the prespecified primary analysis, and previous AI/robot nursing education showed a similar association in supplementary, exploratory analyses. General AI attitudes were measured separately from, and should not be equated with, acceptance or actual use of the specific robotic system. These cross-sectional associations should not be interpreted as evidence of causal effects, objective workload improvement, or readiness for sustained clinical implementation.

  • Open Access

    Hesaplamalı Zekanın Kuramsal Temelleri: Yapay Zeka, Öğrenme Kuramı ve Büyük Veri Paradigması

    by Sara Naghib Zadeh, Zühre Aydın, Hatice Nur Gök, Cansu Arslan, Uygar Aydın, İnCi Zaim Gökbay, Mehmet Dinçer Erbaş, Kenan Peker, Gökhan Önder Ergüven, Tevfik Erdal Baylay, et al.

    Yapay zeka, son yetmiş yıl içerisinde yalnızca bilgisayar bilimlerinin bir alt araştırma alanı olmaktan çıkmış; ekonomi, sağlık, eğitim, mühendislik, güvenlik ve sosyal bilimler başta olmak üzere insan yaşamının hemen her alanını dönüştüren disiplinlerarası bir paradigma haline gelmiştir. Özellikle veri üretim hızındaki artış, hesaplama gücünün gelişmesi ve öğrenme algoritmalarının olgunlaşması, yapay zekâyı teorik bir araştırma alanından günlük yaşamın merkezinde yer alan bir teknolojiye dönüştürmüştür.Ancak yapay zeka alanında yaşanan hızlı gelişmeler, çoğu zaman bu teknolojilerin altında yatan kuramsal temellerin göz ardı edilmesine neden olmaktadır. Oysa günümüzde kullanılan makine öğrenmesi, derin öğrenme, büyük veri analitiği ve karar destek sistemleri; matematiksel modelleme, hesaplama kuramı, istatistiksel öğrenme ve optimizasyon gibi köklü bilimsel birikimlerin üzerine inşa edilmiştir. Bu nedenle yapay zekayı yalnızca uygulama boyutuyla değerlendirmek, alanın bütüncül olarak anlaşılmasını engelleyen önemli bir eksiklik oluşturmaktadır.Çalışma, yapay zeka, öğrenme kuramı ve büyük veri paradigmasının kuramsal temellerini, güncel uygulama alanlarıyla birlikte ele alan disiplinlerarası bir başvuru kaynağı olarak hazırlanmıştır. Kitapta, öğrenen sistemlerin matematiksel ve hesaplamalı altyapısı incelenirken; sağlık, siber güvenlik, sosyal medya analitiği, çok-kipli veri işleme ve veri kıtlığı altında öğrenme gibi güncel araştırma alanlarında geliştirilen yöntemler de ayrıntılı biçimde değerlendirilmektedir. Böylece yapay zekâ, yalnızca algoritmik bir yaklaşım olarak değil, büyük ölçekli veriden bilgi üreten ve farklı disiplinlerde dönüşüm yaratan bütüncül bir bilimsel paradigma olarak ele alınmaktadır. Nitekim büyük veri paradigması, günümüz yapay zeka sistemlerinin gelişimini hızlandıran ve öğrenme süreçlerini yeniden tanımlayan temel itici güçlerden biri haline gelmiştir.Bu eser, yalnızca yapay zeka teknolojilerini kullanan bireylere değil; bu teknolojilerin neden ve nasıl çalıştığını anlamak isteyen araştırmacılara, lisans ve lisansüstü öğrencilerine, akademisyenlere ve sektör profesyonellerine de rehberlik etmeyi amaçlamaktadır. Kitabın temel hedefi, okuyucunun yapay zekâya ilişkin kavramsal ve kuramsal altyapısını güçlendirmek, güncel gelişmeleri daha sağlıklı yorumlayabilmesini sağlamak ve gelecekte ortaya çıkacak teknolojik dönüşümlere bilimsel bir perspektifle yaklaşmasına katkıda bulunmaktır.Yapay zekanın giderek daha fazla otonom karar aldığı, öğrenebildiği ve insan yaşamını şekillendirdiği bir dönemde, bu alanın kuramsal temellerini anlamanın yalnızca akademik bir gereklilik değil, aynı zamanda toplumsal bir sorumluluk olduğu düşünülmektedir. Bu kitabın, yapay zeka, öğrenme kuramı ve büyük veri alanlarında çalışan araştırmacılar, öğrenciler ve sektör profesyonelleri için güvenilir bir başvuru kaynağı olması ve gelecekteki disiplinlerarası araştırmalara ilham vermesi amaçlanmaktadır.

  • Open Access

    Büyük Dil Modellerinin (LLM) Kuramsal Sınırları ve Varsayımları

    by Tevfik Erdal Baylay, Atınç Yılmaz

    Bu bölüm, büyük dil modellerinin kuramsal sınırlarını ve varsayımlarını, sistem başarısızlığı göstergesi değil, kullanım bağlamının sınırlarını tanımlayan özellikler olarak ele alır. Benchmark skorlarının anlamı hangi varsayımların devrede olduğuna bağlıdır; sınır analizi bu varsayım kümesini ve dolayısıyla karar veya bilgi kaynağı olarak modelin hangi bağlamlarda konumlandırılabileceğini görünür kılar. Dilin olasılıksal temsili, sonlu bağlam penceresi ve token düzeyinde eğitim hedefi hesaplama ve temsil varsayımlarını oluşturur; genelleme eğitim–değerlendirme dağılımları ve görev formatına bağımlıdır. Çıktının epistemik güvenilirliği bağlama bağımlıdır; dağıtım koşulları ve doğrulama pratikleri tasarım varsayımlarıyla örtüştüğünde çıktı bilgi kaynağı işlevi görebilir, aksi durumda güvenilirlik ve belirsizlik uygulama katmanında yönetilir. Halüsinasyon tasarım özelliği olarak, hangi görevlerde retrieval, doğrulama modülleri veya insan-in-the-loop ile sınırlandırılması gerektiğinin gerekçesini oluşturur. Açıklanabilirlik ve agent mimarileri, sınırların pratikte nasıl yönetileceği ve hangi mimari güvencelerin devreye alınacağı sorusunu gündeme getirir. Bölüm, LLM etkinliğinin bağlama bağımlı olduğunu ve retrieval, insan denetimi ve modüler doğrulama katmanlarıyla artırılabileceğini vurgulayarak sonuçlandırır.

  • Open Access

    A Symmetry-Aware Adaptive Hybrid Learning Framework with Physics-Informed Representation for Robust Prediction of Concrete Compressive Strength: Proposed ASAPH Framework

    by Atınç Yılmaz, Osman Çaylı

    Accurate prediction of concrete compressive strength remains a challenging problem due to the complex and nonlinear interactions among mixture components and curing conditions. While machine learning approaches have shown promising results, existing studies are typically limited by static model integration strategies and insufficient consideration of structural relationships among input variables. To address these limitations, this study proposes a novel Adaptive Symmetry-Aware Physics-Informed Hybrid (ASAPH) learning framework. The proposed approach integrates three key components: (i) symmetry-consistent feature representation that preserves invariant relationships among mixture parameters, (ii) a stability-driven feature selection mechanism with a relevance–redundancy trade-off, and (iii) an adaptive input-dependent ensemble strategy that dynamically combines multiple learners. In contrast to conventional stacking methods, the proposed framework employs a learnable weighting function to adjust model contributions based on input characteristics, enabling more flexible, robust, and input-adaptive predictions. The framework combines an attention-based tabular model (TabNet) for representation learning and a tree-based ensemble model (XGBoost) for predictive robustness within a unified adaptive fusion architecture. Experimental results on a benchmark dataset using 10-fold cross-validation demonstrate that the proposed model achieves strong predictive performance, with R2 = 0.9162, RMSE = 4.8271, and MAE = 3.4118, outperforming strong baseline models including XGBoost and TabNet. Furthermore, explainability analysis based on SHAP reveals that curing age, cement content, and water-related parameters are the most influential factors governing compressive strength, consistent with established engineering knowledge. Among these, curing age emerges as the most dominant factor, followed by water-related ratios and cement content, indicating strong alignment with established domain knowledge. These findings confirm that incorporating symmetry-aware and physics-informed representations enhances both interpretability and predictive reliability. Overall, the proposed framework provides a principled and generalizable approach for modeling complex engineering systems, bridging the gap between data-driven learning and physically consistent modeling.


  • Open Access

    From selection to first music release: Platformized artist development in the Big5–Manifest case

    by Atınç Yılmaz, Bülent Kabaş

    This study examined how digital platforms participate in the organization of artist development before a group’s first music release. Its case was Big5, a YouTube-centered talent program that publicly presented the selection, training, evaluation, and performance processes leading to the formation of the Turkish girl group Manifest. Using an embedded mixed-methods single-case design, the study integrated documentary sources, a complete inventory of 832 eligible official uploads within a predefined observation window, a blind secondary check of 132 content records, two Google Trends comparisons screened for semantic fit, and multi-label analysis of 1,200 public YouTube comments. The inventory and descriptive content coding showed that 545 uploads (65.5%) appeared before Manifest’s first release and that the group entered the market within a pre-existing content architecture linking formation, performance, participation, and promotion. Official publishing displayed structural continuity, relative search interest aligned with transition and release events, and the exploratory comment analysis identified limited explicit retrospective continuity. The article conceptualizes this arrangement as platformized artist development: selected developmental functions become recurring public content before and across first release, creating conditions for audience familiarization without demonstrating individual audience transfer or established fandom.

  • Open Access

    Evaluation of the Effectiveness of Deep Learning Model in Detection and Classification of Pressure Injury

    by Hamiyet Kızıl, Atınç Yılmaz, Melek Demiral, Umut Kaya, Rıdvan Çakır

    Objective: The study was conducted to determine the effect of the deep learning model on the knowledge and satisfaction levels of nurses in the detection and classification of pressure injuries. Method: The population of this randomized controlled trial consisted of nurses working in intensive care, internal medicine, and surgical clinics at a foundation university hospital between March and April 2022 who voluntarily participated in the study. The sample consisted of a total of 60 (30 experimental and 30 control) nurses who met the sample criteria. The research data were collected using the Structured Nurse Introduction Form, Modified Pieper Pressure Injury Knowledge Test and Nurse Satisfaction Scale.The research data were analyzed in the SPSS 25.0 program. Results: The mean age of the nurses in the experimental group was determined as 25.67±7.27, and the control group as 25.10±3.47. 50% of the nurses in the experimental and control groups graduated from health vocational high schools, and 40% of them worked in surgical services. When the nurses' post-training knowledge exam (post-test) scores were compared; the mean score of the experimental group was determined as 39.36±1.88 and the control group as 33.30±1.68. The post-training knowledge level of the experimental group was found to be statistically significantly higher than the control group (P

  • Open Access

    A Novel Hybrid Deep Learning–Probabilistic Framework for Real-Time Crash Detection from Monocular Traffic Video

    by Reşat Buğra Erkartal, Atınç Yılmaz

    The rapid evolution of autonomous vehicle technologies has amplified the need for crash detection that operates robustly under complex traffic conditions with minimal latency. We propose a hybrid temporal hierarchy that augments a Region-based Convolutional Neural Network (R-CNN) with an adaptive time-variant Kalman filter (with total-variation prior), a Hidden Markov Model (HMM) for state stabilization, and a lightweight Artificial Neural Network (ANN) for learned temporal refinement, enabling real-time crash detection from monocular video. Evaluated on simulated traffic in CARLA and real-world driving in Istanbul, the full temporal stack achieves the best precision–recall balance, yielding 83.47% F1 offline and 82.57% in real time (corresponding to 94.5% and 91.2% detection accuracy, respectively). Ablations are consistent and interpretable: removing the HMM reduces F1 by 1.85–2.16 percentage points (pp), whereas removing the ANN has a larger impact of 2.94–4.58 pp, indicating that the ANN provides the largest marginal gains—especially under real-time constraints. The transition from offline to real time incurs a modest overall loss (−0.90 pp F1), driven more by recall than precision. Compared to strong single-frame baselines, YOLOv10 attains 82.16% F1 and a real-time Transformer detector reaches 82.41% F1, while our full temporal stack remains slightly ahead in real time and offers a more favorable precision–recall trade-off. Notably, integrating the ANN into the HMM-based pipeline improves accuracy by 2.2%, while the time-variant Kalman configuration reduces detection lag by approximately 0.5 s—an improvement that directly addresses the human reaction time gap. Under identical conditions, the best RCNN-based configuration yields AP@0.50 ≈ 0.79 with an end-to-end latency of 119 ± 21 ms per frame (~8–9 FPS). Overall, coupling deep learning with probabilistic reasoning yields additive temporal benefits and advances deployable, camera-only crash detection that is cost-efficient and scalable for intelligent transportation systems.


  • Open Access

    Smart Classroom Scheduling and Resource Optimization for Educational Institutions: Integrating AI and Multi-Objective Decision Support

    by Osman Çaylı, Atınç Yılmaz

    The demand for high-quality education delivery in increasingly dynamic and competitive educational markets has intensified the need for intelligent and adaptive scheduling systems. Manual classroom scheduling methods, which rely on human decision-making, often fail to optimize critical resources such as classrooms, teachers, and student time, leading to inefficiencies and economic losses. This paper proposes a comprehensive Smart Classroom Scheduling and Optimization System (LMSOPT) that leverages Artificial Intelligence (AI), advanced time-series forecasting, constraint-based multi-objective optimization, and real-time data integration. The proposed system employs Long Short-Term Memory (LSTM) neural networks for highly accurate demand forecasting, alongside heuristic and metaheuristic optimization algorithms such as Constraint Programming (CP), Genetic Algorithms (GA), and Tabu Search. The system aims to dynamically balance multiple conflicting objectives: maximizing classroom occupancy rates, minimizing student waiting times, and aligning teacher availability with student preferences. The expected contributions are multifold: significant operational cost savings, measurable improvements in resource utilization, increased student satisfaction, and the creation of an extensible research framework for AI applications in education management. The study aligns with national strategies for digital transformation and supports the vision of data-driven decision-making in educational administration. Empirical results and comparative analyses are presented to validate the system’s effectiveness and demonstrate its replicability for institutions of various scales.


  • Open Access

    A Novel Deep Hybrid Model for Automatic Femoral Stem Classification in Hip Arthroplasty From Radiographs: MSFT-Net With CBAM and Transformer Modules

    by Emre Gogus, Atinç Yilmaz, Meric Enercan

    Accurate identification of femoral stem implants in hip arthroplasty is essential for effective revision surgery, minimizing operative complexity, patient morbidity, intraoperative blood loss, and postoperative recovery time. In cases where prior implant data are unavailable, manual identification is often required, posing significant challenges due to its time-consuming and error-prone nature. To solve this problem, a novel hybrid deep learning architecture that includes a convolutional block attention module and a swin transformer with multi-scale feature fusion from pre-trained architectures DenseNet201, VGG19, and InceptionV3 under the transfer learning paradigm was proposed in this study. The proposed multi-scale feature transformer network was trained and validated on a dataset comprising 1266 anteroposterior (A.P.) hip radiographs of 10 different femoral stem implant types. The proposed hybrid deep learning architecture achieved a training accuracy of 96.7% and validation accuracy of 94.86%, significantly outperforming other baseline models. Compared to state-of-the-art methods, the proposed model achieved an absolute accuracy improvement of 9.5% over VGG19 and 7.4% over DenseNet201 and 8.8% over InceptionV3, demonstrating a significant advancement over existing models in femoral stem classification. The average inference time per image was under 1 second. The experimental results demonstrated that the proposed architecture enhances classification performance while reducing overfitting through attention and transformer-based feature refinement. This automated approach facilitates real-time preoperative implant recognition, thereby streamlining surgical planning, potentially reducing operative costs and duration, and improving clinical outcomes.