Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Thanh-Tuan Nguyen has a total of 29 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Composition-dependent deformation pathways and non-monotonic strength-toughness synergy in CuxCoCrFeNi high-entropy alloy nanowires

    by Xuan-Huy Nguyen, Thanh-Tuan Nguyen, Chau-Tan-Phat Le

    We report statistically replicated molecular dynamics simulations of CuxCoCrFeNi (x = 0.5, 1.0, 1.5) nanowires under uniaxial tension at 100, 300, 600 and 900 K, using the Farkas-Caro embedded-atom method potential with n = 3-4 independent configurations per condition (CVUTS < 2.1% in all 12 cases). Seven mechanical parameters extracted from each stress-strain curve are read against Von Mises shear strain maps taken at three stages of deformation. The equiatomic composition exhibits a non-monotonic strength-toughness optimum. Cu1.0CoCrFeNi at 100 K records both the highest ultimate tensile strength (22.84 ± 0.37 GPa) and the highest toughness modulus (2.06 ± 0.09 GPa) of any condition studied, exceeding Cu0.5 in toughness by deltaUt = 0.20 GPa at essentially the same peak stress. The strain maps at 300 K identify three distinct failure routes: diffuse multi-zone slip in Cu0.5, a single shear band that forms abruptly in Cu1.0, and early planar fracture seeded by void nucleation in Cu1.5. The timing of plastic localisation differs across compositions. Cu1.0 remains in the low-strain regime at both the sub-UTS and near-UTS frames, whereas Cu0.5 already carries slip traces and Cu1.5 shows strain concentrating at two sites. This constitutes direct atomistic evidence for a delayed-localisation mechanism. Independently of composition, the 0.2% offset yield strength falls from roughly 7.5-8.5 GPa at 100 K to 2.2-2.6 GPa at 900 K while the hardening capacity (σUTS -σy)/σUTS rises from about 62% to about 85%, a shift attributed to thermally activated, surface-dominated dislocation nucleation. These results map the deformation pathways of CuxCoCrFeNi nanowires and provide quantitative guidance for designing nanostructured HEAs.

  • Open Access

    TF-SAGE: Trust Filtered Graph Learning for Stable Internet of Things Intrusion Detection under Adversarial Attacks

    by Chin-Shiuh Shieh, Thanh-Lam Nguyen, Thanh-Tuan Nguyen, Xuan-Huy Nguyen, Chau-Tan-Phat Le, Mong-Fong Horng

    Internet of Things (IoT) intrusion detection systems face increasing pressure from adversarial attacks that can manipulate not only feature vectors but also the relational structure on which graph based models rely. This paper proposes Trust Filtered GraphSAGE (TF-SAGE), a graph based intrusion detection system (IDS) pipeline in which edges are assigned trust scores, filtered before message passing, and coupled with uncertainty aware inference to reduce overconfident decisions under unstable neighborhoods. The model is evaluated on NF-ToN-IoT-v2 as the main benchmark and CICIIoT2025 as an independent confirmation benchmark under the same FSAA and GSAA evaluation protocol. The results show that TF-SAGE is not the top clean score model, yet it maintains substantially stronger stability under attack: on NF-ToN-IoT-v2, it reaches clean macro averaged F1 (Macro-F1) 0.9789, retains 0.9776 under Feature Space Adversarial Attack (FSAA), and achieves 0.9048 with attack success rate (ASR) 0.0941 under GSAA, while the graph baselines degrade more severely. Evidence from calibration and neighborhood recovery further indicates that these gains are mechanistically grounded rather than reducible to a single summary score. These findings position TF-SAGE as a practical resilience oriented design direction for IoT intrusion detection based on graph neural networks (GNNs).

  • Open Access

    Real-Time Mouth State Detection Based on a BiGRU-CLPSO Hybrid Model with Facial Landmark Detection for Healthcare Monitoring Applications

    by Mong-Fong Horng, Thanh-Lam Nguyen, Thanh-Tuan Nguyen, Chin-Shiuh Shieh, Lan-Yuen Guo, Chen-Fu Hung, Chun-Chih Lo

    The global population is rapidly expanding, driving an increasing demand for intelligent healthcare systems. Artificial intelligence (AI) applications in remote patient monitoring and diagnosis have achieved remarkable progress and are emerging as a major development trend. Among these applications, mouth motion tracking and mouth-state detection represent an important direction, providing valuable support for diagnosing neuromuscular disorders such as dysphagia, Bell’s palsy, and Parkinson’s disease. In this study, we focus on developing a real-time system capable of monitoring and detecting mouth state that can be efficiently deployed on edge devices. The proposed system integrates the Facial Landmark Detection technique with an optimized model combining a Bidirectional Gated Recurrent Unit (BiGRU) and Comprehensive Learning Particle Swarm Optimization (CLPSO). We conducted a comprehensive comparison and evaluation of the proposed model against several traditional models using multiple performance metrics, including accuracy, precision, recall, F1-score, cosine similarity, ROC–AUC, and the precision–recall curve. The proposed method achieved an impressive accuracy of 96.57% with an excellent precision of 98.25% on our self-collected dataset, outperforming traditional models and related works in the same field. These findings highlight the potential of the proposed approach for implementation in real-time patient monitoring systems, contributing to improved diagnostic accuracy and supporting healthcare professionals in patient treatment and care.

  • Open Access

    Unknown DDoS Attack Detection with Sliced Iterative Normalizing Flows Technique

    by Chin-Shiuh Shieh, Thanh-Lam Nguyen, Thanh-Tuan Nguyen, Mong-Fong Horng

    DDoS attacks represent one of the most pervasive and evolving threats in cybersecurity, capable of crippling critical infrastructures and disrupting services globally. As networks continue to expand and threats become more so... | Find, read and cite all the research you need on Tech Science Press


  • Open Access

    Open-Set Recognition in Unknown DDoS Attacks Detection With Reciprocal Points Learning

    by Chin-Shiuh Shieh, Fu-An Ho, Mong-Fong Horng, Thanh-Tuan Nguyen, Prasun Chakrabarti

    The internet, a cornerstone of modern life, has profound implications across personal, business, and society. However, its widespread use has posed challenges, especially concerning privacy and cybersecurity. Besides, the threats on the internet are increasing in terms of danger, intensity, and complexity. Distributed denial-of-service (DDoS) attacks have emerged as a common and dangerous cybersecurity threat capable of disabling the network systems of targeted organizations and services. Therefore, various security strategies, such as firewalls and intrusion detection systems (IDS), are employed to protect against DDoS attacks. Enhancing the defensive capabilities of IDS systems through machine learning (ML) and deep learning (DL) technologies is a significant trend nowadays. However, despite notable successes, detecting DDoS attacks using ML and DL technologies still faces challenges, especially with Unknown DDoS Attacks. In this research, the primary goal is to address the unknown DDoS detection problem through efficient and advanced techniques. Our proposed method, CNN-RPL, integrates Convolutional Neural Network (CNN) with Reciprocal Points Learning (RPL), a novel Open-Set Recognition (OSR) technology. This model can effectively handle both known and unknown attacks. The CNN-RPL model demonstrates excellent results, achieving an accuracy exceeding 99.93% against known attacks in the CICIDS2017 dataset. Simultaneously, the model achieves a commendable average accuracy of up to 98.51% against unknown attacks in the CICDDoS2019 dataset. In particular, the CNN-RPL model simplifies the architecture of the deep neural network by significantly reducing the number of training parameters without compromising defense capabilities. Therefore, our proposed method is genuinely efficient, particularly flexible, and lightweight compared to traditional methods. This can equip organizations and businesses with a highly applicable yet powerful security approach against the evolving complexities in the network space.


  • Unknown DDoS Attack Detection Using Open-Set Recognition Technology and Fuzzy C-Means Clustering

    by Hao Kao, Thanh-Tuan Nguyen, Chin-Shiuh Shieh, Mong-Fong Horng, Lee Yu Xian, Denis Miu


  • Reciprocal Points Learning Based Unknown DDoS Attacks Detection

    by Fu-An Ho, Chin-Shiuh Shieh, Mong-Fong Horng, Thanh-Tuan Nguyen, Ying-Chieh Chao

  • Open Access

    Detection of Unknown DDoS Attack Using Convolutional Neural Networks Featuring Geometrical Metric

    by Chin-Shiuh Shieh, Thanh-Tuan Nguyen, Mong-Fong Horng

    DDoS attacks remain a persistent cybersecurity threat, blocking services to legitimate users and causing significant damage to reputation, finances, and potential customers. For the detection of DDoS attacks, machine learning techniques such as supervised learning have been extensively employed, but their effectiveness declines when the framework confronts patterns exterior to the dataset. In addition, DDoS attack schemes continue to improve, rendering conventional data model-based training ineffectual. We have developed a novelty open-set recognition framework for DDoS attack detection to overcome the challenges of traditional methods. Our framework is built on a Convolutional Neural Network (CNN) construction featuring geometrical metric (CNN-Geo), which utilizes deep learning techniques to enhance accuracy. In addition, we have integrated an incremental learning module that can efficiently incorporate novel unknown traffic identified by telecommunication experts through the monitoring process. This unique approach provides an effective solution for identifying and alleviating DDoS. The module continuously improves the model’s performance by incorporating new knowledge and adapting to new attack patterns. The proposed model can detect unknown DDoS attacks with a detection rate of over 99% on conventional attacks from CICIDS2017. The model’s accuracy is further enhanced by 99.8% toward unknown attacks with the open datasets CICDDoS2019.


  • Open Access

    Detection of Unknown DDoS Attack Using Reconstruct Error and One-Class SVM Featuring Stochastic Gradient Descent

    by Chin-Shiuh Shieh, Thanh-Tuan Nguyen, Chun-Yueh Chen, Mong-Fong Horng

    The network system has become an indispensable component of modern infrastructure. DDoS attacks and their variants remain a potential and persistent cybersecurity threat. DDoS attacks block services to legitimate users by incorporating large amounts of malicious traffic in a short period or depleting system resources through methods specific to each client, causing the victim to lose reputation, finances, and potential customers. With the advancement and maturation of artificial intelligence technology, machine learning and deep learning are widely used to detect DDoS attacks with significant success. However, traditional supervised machine learning must depend on the categorized training sets, so the recognition rate plummets when the model encounters patterns outside the dataset. In addition, DDoS attack techniques continue to evolve, rendering training based on conventional data models unable to meet contemporary requirements. Since closed-set classifiers have excellent performance in cybersecurity and are quite mature, this study will investigate the identification of open-set recognition issues where the attack pattern does not accommodate the distribution learned by the model. This research proposes a framework that uses reconstruction error and distributes hidden layer characteristics to detect unknown DDoS attacks. This study will employ deep hierarchical reconstruction nets (DHRNet) architecture and reimplement it with a 1D integrated neural network employing loss function combined with spatial location constraint prototype loss (SLCPL) as a solution for open-set risks. At the output, a one-class SVM (one-class support vector machine) based on a random gradient descent approximation is used to recognize the unknown patterns in the subsequent stage. The model achieves an impressive detection rate of more than 99% in testing. Furthermore, the incremental learning module utilizing unknown traffic labeled by telecom technicians during tracking has enhanced the model’s performance by 99.8% against unknown threats based on the CICIDS2017 Friday open dataset.