Yazar "Asal, Burcak" seçeneğine göre listele

Listeleniyor 1 - 7 / 7
  • [ X ]
    Öğe
    Adaptive Video Anomaly Detection by Attention-Based Relational Knowledge Distillation
    (IEEE-Inst Electrical Electronics Engineers Inc, 2025) Asal, Burcak; Can, Ahmet Burak
    Detecting anomaly patterns in videos is a challenging task due to complex scenes, huge diversity of anomalies, and fuzzy nature of the task. With advent of technology, tremendous size of visual data is being generated by video surveillance systems, which makes harder to search, analyze, and detect anomalies on video data by human operators. In this paper, we introduce three relational distillation approaches to handle both robust detection of anomalous events and gradual adaptation to different anomaly patterns in new videos while not forgetting anomaly patterns learned from the previous video data. In order to realize these concepts, we propose a unique attention mechanism with feature and relation based knowledge distillation methods. We adapted our knowledge distillation methods to two state-of-the-art models designed for anomaly detection task. Our extensive experiments on two public datasets show that not only our best version model achieves robust performance with a frame-level AUC of 80.22 on UCF-Crime and video-level AUC of 78.20 on RWF-2000 datasets but also the proposed distillation methods improve the performance while reducing catastrophic forgetting problem.
  • [ X ]
    Öğe
    Benchmarking TabNet, NODE, and FT-Transformer for Software Defect Prediction: An Empirical Comparison and Explainability Analysis
    (IEEE-Inst Electrical Electronics Engineers Inc, 2026) Asal, Burcak; Yalciner, Burcu
    Software defect prediction (SDP) is essential for improving software quality and reliability. Traditional machine learning methods, while effective, often fail in capturing complex interactions among software metrics. Recently, specialized deep learning architectures designed for tabular data, including TabNet, Neural Oblivious Decision Ensembles (NODE), and FT-Transformer, have emerged, offering promising potential to enhance prediction accuracy and interpretability. This study comprehensively benchmarks the TabNet, NODE and FT-Transformer models on the challenging NASA JM1 dataset from the PROMISE repository. We address severe class imbalance using NearMiss undersampling and ensure hyperparameter optimization for fairness across comparisons. The performance of the models was evaluated using standard metrics: precision, recall, F1-score, and accuracy. In addition, the interpretability of the model was assessed using SHAP and LIME methods. The FT-Transformer and NODE models demonstrated superior performance, achieving 88% accuracy compared to the accuracy of TabNet 86%. FT-Transformer showed exceptional precision (99%) for defect detection, emphasizing its low false-positive rate. SHAP and LIME analyzes revealed unique attention patterns for each model, highlighting differences in feature importance and decision-making processes. FT-Transformer and NODE outperform TabNet in accuracy and balance between recall and precision. Interpretability analysis provides actionable insights into feature importance, enabling better decision-making in practical SDP workflows.
  • [ X ]
    Öğe
    Deep learning and explainable AI for email phishing classification: A comparative study of TabNet, NODE and FT-transformer models
    (Gazi Univ, 2025) Asal, Burcak; Oyucu, Saadin; Dogan, Ferdi; Polat, Onur; Aksoz, Ahmet
    In the changing landscape of cybersecurity threats, phishing emails indicate a persistent and damaging attack vector. This study investigates the effectiveness of deep learning models on a phishing email classification task using tabular data and focusing on TabNet, NODE (Neural Oblivious Decision Ensembles), and FT-Transformer architectures. The utilized dataset includes eight input features capturing linguistic and structural characteristics of emails, with a binary label indicating phishing or normal classification. Additionally, the NearMiss under-sampling approach is applied to address the significant class imbalance. Experimental results demonstrate that while all three models achieve strong performance, the FT-Transformer model outperforms TabNet and NODE by achieving the highest classification accuracy and balanced precision-recall scores. Additionally, explainable artificial intelligence (XAI) methods, SHAP and LIME, are employed to interpret the FT-Transformer model's decision-making process, which highlights the critical role of spelling errors, unique word counts, and urgency-related keywords in phishing detection. The findings emphasize the potential of transformer-based approaches for tabular cybersecurity applications and indicate the importance of interpretable AI in enhancing trust and transparency in phishing detection systems.
  • [ X ]
    Öğe
    Electrification in Maritime Vessels: Reviewing Storage Solutions and Long-Term Energy Management
    (MDPI, 2025) Aksoz, Ahmet; Asal, Burcak; Golestan, Saeed; Gencturk, Merve; Oyucu, Saadin; Bicer, Emre
    Electric and hybrid marine vessels are marking a new phase of eco-friendly maritime transport, combining electricity and traditional propulsion to boost efficiency and reduce emissions. The industry's advancements in charging infrastructure and strict regulations help these vessels lead the way toward a sustainable and economically viable future in shipping. In this review, electric and hybrid marine vessels are discussed, including past applications and trend demonstrations. This paper systematically analyzes maritime vessels' energy management and battery systems, highlighting advances in lithium-based and alternative battery technologies. Additionally, the review examines the impact of these technologies on sustainability and operational efficiency in the maritime industry. This paper contributes to the field by presenting a holistic view of the challenges and solutions associated with the electrification of maritime vessels, aiming to inform future developments and policymaking in this dynamic sector. Unlike many existing reviews that focus exclusively on battery chemistries or energy management algorithms, this manuscript integrates multiple aspects of maritime electrification-including propulsion types, charging infrastructure, grid systems (MVDC), EMS, BMS, and AI applications-into one cohesive systems-level review. This cross-sectional integration is particularly rare in the literature and enhances the practical value of the review for designers, policymakers, and shipbuilders.
  • [ X ]
    Öğe
    Enhancing Industrial IoT Cybersecurity with Explainable AI: A SHAP and LIME-Based Intrusion Detection Methodology
    (IEEE, 2025) Asal, Burcak; Cakin, Alperen; Dilek, Selma
    Although the proliferation of Industrial Internet of Things (IIoT) systems has transformed industrial operations, it has also introduced significant cybersecurity challenges. Ensuring IIoT network security requires robust, interpretable models capable of detecting and mitigating threats. This study integrates Explainable Artificial Intelligence (XAI) techniques SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-Agnostic Explanations) to enhance the interpretability of machine learning-based intrusion detection in IIoT. Using the WUSTL-IIoT-2021 dataset, we evaluated Conditional Variational Autoencoder (CVAE), Decision Tree (DT), and Random Forest (RF) models, analyzing their transparency and performance. SHAP and LIME identify critical features such as DstJitter, Dport, and SAppBytes, contributing to improved explainability. RF achieves near-perfect accuracy (99.99%), while optimized feature subsets maintain high accuracy with lower computational cost. The results highlight XAI's role in balancing accuracy, interpretability, and efficiency in IIoT cybersecurity, paving the way for more trustworthy intrusion detection systems.
  • [ X ]
    Öğe
    Interpretable Deep Learning for Pulsar Star Classification with Explainable AI Techniques: A Comparative Analysis of TabNet and FT-Transformer Models
    (IEEE, 2025) Asal, Burcak
    The classification of pulsar stars from vast astronomical datasets is an important task in astrophysics, which aids in the discovery of these rapidly rotating neutron stars that emit periodic radio signals. Traditional machine learning techniques have demonstrated effectiveness in pulsar detection but often can lack the capacity to handle complex feature interactions within structured data. Deep learning models, especially designed for tabular data, have emerged as promising alternatives. In this study, TabNet and FT-Transformer, two advanced deep learning architectures specifically designed for tabular data are explored to enhance pulsar star classification on the HTRU2 dataset. Additionally, to improve model transparency, two explainable artificial intelligence (XAI) techniques, which are SHAP (SHapley Additive Explanations) and LIME (Local Interpretable Model-Agnostic Explanations), are integrated to analysis both global and local feature contributions in the classification process.
  • [ X ]
    Öğe
    Interpretable Energy Forecasting: Comparative Analysis of Voting Regression and NODE Models for Electricity Power Consumption Prediction
    (IEEE, 2025) Asal, Burcak
    Accurate electricity power consumption estimation is vital for efficient energy management, demand-side response strategies and grid optimization. This study explores ensemble based machine learning-based and deep learning approaches for electricity power consumption prediction across three distinct zones by using a Voting Regression model including Linear Regression, Random Forest and K-Nearest Neighbors (KNN) and NODE (Neural Oblivious Decision Ensembles) model. Quantitative and qualitative experimental results show that Voting Regression model outperforms NODE model in overall and additionally, SHAP (SHapley Additive exPlanations) technique is applied on the Voting Regression model to analyze and explain the feature contributions and effects for the model prediction. Consequently, this study provides a reliable, explainable and data-driven framework for optimizing electricity power distribution and electricity consumption planning.