http://dspace.aiub.edu:8080/jspui/handle/123456789/55 Publications Authored By All Faculty Members of Engineering (Department of Architecture, EEE and IP) Thu, 27 Aug 2026 19:41:13 GMT 2026-08-27T19:41:13Z http://dspace.aiub.edu:8080/jspui/handle/123456789/2987 Title: An End-to-End Framework for Fair and Transparent Decision Support using Explainable Deep Learning Authors: Mustavy, Md. Ridwan Al; Jahan, Marowa; Partho, Md. Mobashir Tajuare; Uddin, Md. Helal; Bhuyan, Muhibul Haque Abstract: This paper introduces FairXDL, a decision support system that jointly maximizes prediction accuracy and demographic fairness via a differentiable fairness regularizer. Leveraging the power of a compact residual deep neural network architecture, FairXDL uses Integrated Gradients to produce human-readable interpretations of individual samples without any additional forward passes. In experiments on the COMPAS recidivism dataset ($n = 7,214$), FairXDL increases the demographic parity ratio by 18.4 percentage points while suffering no more than a 0.9% reduction in prediction accuracy compared to a fairness-naive model. A streamlined ONNX export pipeline verifies the feasibility of deploying this approach in production. Description: Self-funded research. Fri, 19 Jun 2026 00:00:00 GMT http://dspace.aiub.edu:8080/jspui/handle/123456789/2987 2026-06-19T00:00:00Z http://dspace.aiub.edu:8080/jspui/handle/123456789/2986 Title: Deep learning and quantum-enhanced predictive optimization in power electronics Authors: Jahan, Marowa; Mustavy, Md Ridwan Al; Bhuyan, Md Ridwan Al Mustavy Abstract: Optimization plays a crucial part in the plan, control, and operation of modern power electronic systems. Traditional methods, viz. Genetic Algorithm, Particle Swarm Optimization (PSO), and Differential Evolution have been widely used to optimize converter efficiency, stability, and performance. However, the increasing complexity of renewable energy systems, electric vehicles, and smart grids necessitate advanced optimization frameworks. This paper discovers the incorporation of Artificial Intelligence, Machine Learning, and Quantum Machine Learning into power electronics optimization. Reinforcement Learning is investigated for adaptive control of converters and motor drives, while Neural Networks are explored for predictive control. Hybrid optimization methods, viz Fuzzy with PSO and Artificial Neural Networks with Genetic Algorithm, are presented to improve convergence speed and accuracy. Simulation platforms, like MATLAB and Python are leveraged to evaluate optimization frameworks. Finally, we introduce a novel deep learning-based predictive controller augmented with QML techniques for converters in EV and renewable systems. We propose a DL and QML-based predictive controller that attains around 15 % lower converter losses compared to classical methods. Description: Students and faculty members invested in this research. Tue, 23 Jun 2026 00:00:00 GMT http://dspace.aiub.edu:8080/jspui/handle/123456789/2986 2026-06-23T00:00:00Z http://dspace.aiub.edu:8080/jspui/handle/123456789/2985 Title: GDR-GNN: A Graph Neural Network Framework for Explainable Hereditary Genetic Disorder Risk Prediction Authors: Amit, Arpon Paul; Oni, Saiful Islam; Tanvir, Kazi; Gomes, Dipta Justin; Rahman, Mahfujur; Bhuyan, Muhibul Haque Abstract: Accurate prediction of risk of hereditary genetic disorder is necessary for early intervention and clinical decision making. This study proposes a novel graph based learning model, considering clinical and genetic tabular data to be converted into a relational k nearest neighbor graph. Genetic Disorder Risk Graph Neural Network (GDR-GNN), to understand the inter-sample dependency. Automatic hyper-parameter optimization Optuna to perform to search the parameter space and find the optimal GNN configuration besides, on our sequence, and explainable AI techniques (SHAP and LIME) are integrated to provide both global and local interpretability mechanisms. The model is evaluated by testing it on a clinically validated dataset of 100 families and comparing it to the best- performing machine learning classifiers, including SVM, KNN, Random Forest, SVM, AdaBoost and Histogram based gradient boosting. The experiment outcomes show that the proposed GDR-GNN can achieve better predictive accuracy than the traditional models with the accuracy of 97.67 % and Cohen kappa of 0.9651, which is superior to the traditional models and provides greater transparency. The XAI analyses indicate that the model is based around clinical and genetic attributes such as parental carrier status, consanguinity, and the range of specific gene values, which have a central role in the model, and supports its clinical reliability. These findings highlight the potential of graph-based modelling and explainability-driven methods in order to advancing hereditary risk prediction for real world clinical use. Description: 10,000 Taka was expended for this research. Conference registration fee of 10,000 Taka was provided by AIUB. Thu, 11 Jun 2026 00:00:00 GMT http://dspace.aiub.edu:8080/jspui/handle/123456789/2985 2026-06-11T00:00:00Z http://dspace.aiub.edu:8080/jspui/handle/123456789/2984 Title: Machine Learning-Assisted Electronic Voting with SHA3-Based Security Authors: Khan, Mursalin; Mia, Jewel; Rahman, Anisur; Azad, Humaira; Bhuyan, Muhibul Haque Abstract: This paper aims to develop a software-based electronic voting system using a camera and computer as hardware. In this context, the paper addresses the electronic voting security issues by proposing a sophisticated system that can gather information from users, encrypt it using the globally recognized hash algorithm known as SHA-3, and validate it through facial recognition. This system is resilient against jamming, hacking, spoofing, data theft, insider threats, malware injection, physical tampering, side-channel attacks, network exploits, firmware manipulation, data extraction via memory dumps, supply chain attacks, lack of encryption, and exploitation of software vulnerabilities. Addressing these concerns is essential for future progress in secure, scalable, and reliable electronic voting platforms. The test results and findings of the proposed system underscore the need for ongoing research to enhance security, privacy, and efficiency in electronic voting technologies. We didn't use any large photo data set and hence we could avoid the cost of space and data training time. Therefore, we didn't need to compute the values of precision and recall values. Instead, we used cv2.CascadeClassifier('data/haarcascade_frontalface_default.x ml') for the face detection. Description: 10,000 Taka was expended for this research. Conference registration fee of 10,000 Taka was provided by AIUB. Thu, 11 Jun 2026 00:00:00 GMT http://dspace.aiub.edu:8080/jspui/handle/123456789/2984 2026-06-11T00:00:00Z