Yılmaz Kılıçaslan

Prof. Dr. Yılmaz Kılıçaslan

Department of Artificial Intelligence and Data Engineering, İstanbul Atlas University, Istanbul, İstanbul, Turkey (Türkiye)
orcid

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
3 November 2025 in The European Research Journal
Verified
Journal Article
Open access
YFYelda Ali FIRAT
Yelda Ali FIRAT
MSMeral Seferoğlu
Meral Seferoğlu
HKHakan Kılıçaslan
Hakan Kılıçaslan
ASAli Özhan Sıvacı
Ali Özhan Sıvacı
MYMurat Kaan Yılmaz
Murat Kaan Yılmaz
YKYılmaz Kılıçaslan
Yılmaz Kılıçaslan
Objectives: Multiple sclerosis (MS) is a chronic central nervous system disorder that causes demyelination, inflammation, and axonal damage, leading to permanent disabilities in motor, sensory, visual, and balance functions. This study aimed to develop an artificial intelligence (AI)-based, non-invasive diagnostic approach for MS detection using handwriting analysis, leveraging deep learning methods to identify disease-specific handwriting patterns. Methods: A classification model was designed using a convolutional neural network (CNN) based on the VGG16 architecture with transfer learning. The dataset consisted of 426 handwriting samples, including 213 from MS patients and 213 from healthy individuals. Data augmentation and early stopping techniques were employed to improve model generalization capability. Results: The proposed model achieved a validation accuracy of 83.72% and a test accuracy of 85%, indicating its robustness in distinguishing MS patients from healthy subjects. The confusion matrix analysis demonstrated a sensitivity of 86% and a specificity of 84%, indicating moderate discriminatory performance. Conclusions: The findings suggest that the developed AI-based model offers an effective, non-invasive diagnostic tool for MS detection. This approach provides a promising foundation for future research on monitoring disease progression and developing clinically applicable AI-supported diagnostic systems.

26 August 2023 in Computer Science
Verified
Journal Article
Closed access
AÇAyşegül ÇAĞLI
Ayşegül ÇAĞLI
VKVakkas Karakurt
Vakkas Karakurt
KYKürşat Edabalı YILDIRIM
Kürşat Edabalı YILDIRIM
FSFatih Soygazi
Fatih Soygazi
YKYılmaz Kılıçaslan
Yılmaz Kılıçaslan

9 September 2026 in Mathematics
Verified
Journal Article
Open access
PDF file
GAGözde Alp
Gözde Alp
FSFatih Soygazi
Fatih Soygazi
YKYılmaz Kılıçaslan
Yılmaz Kılıçaslan
Although population-based metaheuristic algorithms have been widely applied to the One-Dimensional Cutting Stock Problem (1D-CSP), their performance is often limited by premature convergence and insufficient local search capability. This study presents a comparative investigation of the effect of local search on four population-based metaheuristic paradigms for the one-dimensional cutting stock problem (1D-CSP). Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), and Grey Wolf Optimizer (GWO) are evaluated both in their standard forms and after incorporating a common Simulated Annealing (SA)-based local refinement procedure. The objective is not to introduce a new hybridization strategy but to systematically examine whether and to what extent the same local search mechanism affects algorithms with different search characteristics. The methods are evaluated on two complementary benchmark datasets comprising 39 industrial instances from the Japanese chemical fiber industry and 1800 CUTGEN1 instances. Performance was evaluated using clipping loss, computational cost, Friedman ranks, win counts, and paired Wilcoxon signed-rank tests. According to the results, statistically significant differences were observed for GA, PSO, and GWO across both benchmark sets, while the difference between ACO and M-ACO was not statistically significant. The study provides a comparative assessment of how the common local search component affects different metaheuristic paradigms for full 1D-CSP.

8 October 2026 in Biology
Verified
Journal Article
Open access
PDF file
YFYelda Ali FIRAT
Yelda Ali FIRAT
YKYılmaz Kılıçaslan
Yılmaz Kılıçaslan
Transferability of transcriptome-based gene-dependency models from two-dimensional cancer models to organoids remains uncertain. Expression-only Elastic Net models for 20 preselected genes were selected using patient-grouped nested cross-validation in 681 models and evaluated in a locked 171-model holdout. Historical target selection used all 852 models, making holdout performance conditional internal confirmation rather than an unbiased generalization estimate. After refitting on all 852 models, frozen models underwent exploratory evaluation in 94 Broad organoids and blinded validation in 148 Sanger organoids from 144 donors. Predictions were sealed before Sanger outcome access. Primary target-wise Spearman correlations, ranking organoids per gene, averaged 0.4094, 0.1518, and 0.1015 in holdout, Broad, and Sanger, respectively. Sanger association exceeded the within-lineage permutation null (p = 0.00040), but improvement over lineage-only prediction remained uncertain (difference, 0.0462; 95% confidence interval, −0.0065–0.0986). CDK6, MDM4, and CCNE1 passed false discovery rate correction in Sanger. Retrospective within-organoid gene ranking modestly improved over lineage-only prediction in both cohorts; top-three overlap improved only in Sanger. Pancreatic results remained unvalidated projections. Findings support weak cross-domain rank transfer and limited within-organoid ranking within the selected panel, without establishing genome-wide transferability, causal explanations for attenuation, or prospective experimental or therapeutic utility.

28 April 2026 in JUCS - Journal of Universal Computer Science
Verified
Journal Article
Open access
PDF file
YFYelda Ali FIRAT
Yelda Ali FIRAT
YKYılmaz Kılıçaslan
Yılmaz Kılıçaslan
HSHüseyin Ali SARIKAYA
Hüseyin Ali SARIKAYA
MYMurat Kaan Yılmaz
Murat Kaan Yılmaz
Autism Spectrum Disorder is a neurodevelopmental disorder with onset in early childhood and its diagnosis often requires clinical processes based on long, subjective observations. Although early diagnosis and intervention can significantly improve developmental outcomes, existing methods are limited in terms of scalability and objectivity. The aim of this study is to develop a hybrid deep learning model that detects Autism Spectrum Disorder with high accuracy by analyzing motor behaviors from videos of children recorded in their natural home environment. In this study, joint coordinates were extracted using the MediaPipe Pose model and spatial, temporal, frequency and coordination-based features were calculated from these data. The features were processed with a hybrid architecture integrating CNN, BiLSTM and attention mechanism. CNN captured spatial patterns, BiLSTM learned the dynamics over time, and the attention mechanism focused on critical movement segments. The model achieves over 97% accuracy on closed datasets and over 83% on public videos such as YouTube and TikTok. These results show that the method performs robustly under both controlled and real-world conditions. The study provides a scalable, objective and clinically applicable screening tool that overcomes the problems of artificial environments and limited data.