Attention-Based Models for Multivariate Time Series Forecasting: Multi-step Solar Irradiation Prediction Sadman Sakib, Mahin K. Mahadi, Samiur R. Abir, Al-Muzadded Moon, Ahmad Shafiullah, Sanjida Ali, Fahim Faisal, Mirza M. Nishat Heliyon, 2024 Bangladesh's subtropical climate with an abundance of sunlight throughout the greater portion of the year results in increased effectiveness of solar panels. Solar irradiance forecasting is an essential aspect of grid-connected photovoltaic systems to efficiently manage solar power's variation and uncertainty and to assist in balancing power supply and demand. This is why it is essential to forecast solar irradiation accurately. Many meteorological factors influence solar irradiation, which has a high degree of fluctuation and uncertainty. Predicting solar irradiance multiple steps ahead makes it difficult for forecasting models to capture long-term sequential relationships. Attention-based models are widely used in the field of Natural Language Processing for their ability to learn long-term dependencies within sequential data. In this paper, our aim is to present an attention-based model framework for multivariate time series forecasting. Using data from two different locations in Bangladesh with a resolution of 30 min, the Attention-based encoder-decoder, Transformer, and Temporal Fusion Transformer (TFT) models are trained and tested to predict over 24 steps ahead and compared with other forecasting models. According to our findings, adding the attention mechanism significantly increased prediction accuracy and TFT has shown to be more precise than the rest of the algorithms in terms of accuracy and robustness. The obtained mean square error (MSE), the mean absolute error (MAE), and the coefficient of determination (R 2 ) values for TFT are 0.151, 0.212, and 0.815, respectively. In comparison to the benchmark and sequential models (including the Naive, MLP, and Encoder-Decoder models), TFT has a reduction in the MSE and MAE of 8.4–47.9% and 6.1–22.3%, respectively, while R 2 is raised by 2.13–26.16%. The ability to incorporate long-distance dependency increases the predictive power of attention models.
Implementation of Personal Safety Equipment Tracking & Detection by DeepSORT & YOLOv8 Mahin Khan Mahadi, Rummanur Rahad, Md Sadi Mobassir, Asma Rahman, Ahmad Shafiullah, Mirza Muntasir Nishat 7th International Conference on Inventive Computation Technologies Icict 2024, 2024 The purpose of this research is to develop better techniques for detecting the presence of safety gear such as helmets and vests in hazardous jobs like building, mining, and policing. With a fully convolutional neural network and a self-attention mechanism for effective item recognition, YOLOv8, a new object detection model, outperforms its predecessors. Notably, YOLOv8 removes anchors, which reduces box predictions and speeds up post-processing. With an impressive 96.6% mean average precision (mAP) accuracy in our custom dataset, YOLOv8 is useful for monitoring personal protective equipment usage and improving worker safety on construction sites. The DeepSORT algorithm integration adds sophistication, allowing for extensive object-tracking capabilities. This collaborative approach strengthens safety regulations, representing a substantial development in occupational safety in high-risk workplaces.
Understanding Machine Learning & its Application in Obesity Estimation by Explainable AI Mahin Khan Mahadi, Rummanur Rahad, Abdullah, Abu Noman, Samiha Ishrat, Fahim Faisal 7th International Conference on Inventive Computation Technologies Icict 2024, 2024 The escalating prevalence of obesity poses a formidable challenge to global public health, necessitating innovative approaches for accurate estimation and management. This research addresses this imperative by introducing an Explainable Artificial Intelligence (XAI) framework to estimate obesity levels. Meticulous hyperparameter tuning has resulted in elevated performance metrics, including accuracy, weighted precision, weighted recall, and weighted F1-score. Leveraging a decision support system, a robust machine learning model is developed exhibiting an impressive cross-validation accuracy of 97.39%. The model seamlessly integrates data on eating habits and physical condition, demonstrating improved performance in estimating obesity levels. The application of SHAP analysis unveils critical features within the dataset, thereby augmenting model interpretability and trustworthiness. The outcomes of this study provide a reliable tool for physicians, contributing to more informed clinical decisions in obesity management.
Highly Sensitive Optically Tunable Transition Metal Nitride-Based Plasmonic Pressure Sensor With CMOS-Compatibility at Compact Subwavelength Dimensions Rummanur Rahad, Mohammad Ashraful Haque, Mahin Khan Mahadi, Abu S. M. Mohsin, Md. Omar Faruque, Sheikh Mohd. Ta-Seen Afrid, Md. Jahidul Hoq Emon, Rakibul Hasan Sagor IEEE Sensors Journal, 2024 This article presents a novel transition metal nitride (TMN)–based plasmonic pressure sensor (PPS) that utilizes a TMN-Insulator-TMN structure. Our proposed sensor is built with Zirconium Nitride (ZrN), an alternative plasmonic material that provides several benefits over conventional plasmonic metals like silver and gold. Notably, ZrN is compatible with standard Complementary Metal Oxide Semiconductor (CMOS) technology and it offers tunability of optical properties. The sensor demonstrates a maximum pressure sensitivity of 177.56 nm/MPa with a resolution of 5.63 × 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-6</sup> MPa. Furthermore, the incorporation of TMN endows the PPS with several beneficial characteristics, such as exceptional hardness, high-temperature thermal stability, optical tunability, and lower electrical resistivity which are typically absent in conventional plasmonic noble metals, limiting their practical application in plasmonic devices. Consequently, our proposed configuration outperforms the conventional noble material-based Metal-Insulator-Metal (MIM) configuration, which removes the barriers to wide-scale adaptations of plasmonic devices by paving the way for the research and development of efficient, robust, and long-lasting sensors at subwavelength scales, thereby forging a link between nanoelectronics and plasmonics.
A Phishing Detection Approach for Empowering Cybersecurity with Explainable AI and SelectKBest Mahin Khan Mahadi, Rummanur Rahad, Md. Zidan Shahriar, Rahul Debnath, Fyaz Nafin Rahman, A.M.Tamim Haider, Abdullah Al Zayed, Jubaer Al Mahmud 2024 27th International Conference on Computer and Information Technology Iccit 2024 Proceedings, 2024 Phishing websites are a big cyber threat in today’s online world. They pretend to be real websites and steal personal information from users who are not aware of it. With more users accessing the internet quickly and knowing how to use computers, there’s been a rise in these fake sites that look very real and are made to trick people. It’s a challenging task to spot and sort out these fake sites because of their complexity and dissimilarity in nature. Our study looks at how well the SelectKBest feature selection method works to make tree-based classifiers like Random Forest (RF) and ExtraTrees (ETC) better at finding these phishing sites. We found that by picking the best 40 to 60 features, these classifiers could identify phishing sites with an accuracy of 96.68% after we fine-tuned their settings. We also used SHAP(SHapley Additive exPlanations) analysis, which helps us understand which parts of the data are most important, making our models more reliable and trustworthy. Our research is a step towards creating systems that can detect phishing sites in real-time and quickly warn people about them, which will help improve online security.
Explainable Transfer Learning for Precise Alzheimer's Disease Prediction from MRI Data Mahin Khan Mahadi, Jamal Uddin, Md. Zidan Shahriar, Rummanur Rahad, A.M. Tamim Haider, Rashed Hasan Ratul, Ahmad Shafiullah, Asif Newaz, Ashik Ahmed 2024 International Conference on Innovations in Science Engineering and Technology Innovative Technologies for Global Solutions Iciset 2024, 2024
Machine Learning Assisted Decision Support System for Prediction of Prostrate Cancer Mahin Khan Mahadi, Samiur Rashid Abir, Al-Muzadded Moon, Muhammad Adnan, Mohd Abdun Nafee Islam Khan, Mirza Muntasir Nishat, Fahim Faisal, Md Taslim Reza 2023 20th International Conference on Electrical Engineering Electronics Computer Telecommunications and Information Technology Ecti Con 2023, 2023
BSAF-Based Binary Offloading for Edge AI: Energy-Efficient Deep Learning and Spatial Computing in Mixed Reality MK Mahadi, DM Doe, X Li, Z Han, L Qian 2026 International Conference on Computing, Networking and Communications … , 2026 2026
Dual-mode CMOS-compatible optically tunable plasmonic sensor with symmetrical cavity for simultaneous and independent monitoring of pressure and temperature deviations from … R Rahad, JD Joy, MS Rahman, MJH Emon, R Rahad, MK Mahadi Surfaces and Interfaces, 108325 , 2025 2025
Energy-Efficient Task Offloading Frameworks For Mixed Reality And Extended Reality In Edge AI Environments MK Mahadi Prairie View A&M University , 2025 2025
A phishing detection approach for empowering cybersecurity with explainable ai and selectkbest MK Mahadi, R Rahad, MZ Shahriar, R Debnath, FN Rahman, AMT Haider, ... 2024 27th International Conference on Computer and Information Technology … , 2024 2024 Citations: 1
Gated recurrent unit (GRU)-based deep learning method for spectrum estimation and inverse modeling in plasmonic devices MK Mahadi, R Rahad, MA Haque, MM Nishat Applied Physics A 130 (11), 784 , 2024 2024 Citations: 21
RoadSense: a framework for road condition monitoring using sensors and machine learning IA Jahan, AS Huq, MK Mahadi, IA Jamil, MZ Shahriar IEEE Transactions on Intelligent Vehicles , 2024 2024 Citations: 14
Explainable Transfer Learning for Precise Alzheimer’s Disease Prediction from MRI Data MK Mahadi, J Uddin, MZ Shahriar, R Rahad, AMT Haider, RH Ratul, ... 2024 International Conference on Innovations in Science, Engineering and … , 2024 2024 Citations: 1
An alternative plasmonic material-based CMOS-compatible temperature sensor R Rahad, MM Sobhani, MJH Emon, SMTS Afrid, MK Mahadi, ASM Mohsin, ... Optics Communications 569, 130749 , 2024 2024 Citations: 63
A polarization independent highly sensitive metasurface-based biosensor for lab-on-chip applications R Rahad, MA Haque, MK Mahadi, MO Faruque, SMTS Afrid, ASM Mohsin, ... Measurement 231, 114652 , 2024 2024 Citations: 98
Highly sensitive optically tunable transition metal nitride-based plasmonic pressure sensor with CMOS-compatibility at compact subwavelength dimensions R Rahad, MA Haque, MK Mahadi, ASM Mohsin, MO Faruque, SMTS Afrid, ... IEEE sensors journal 24 (14), 22271-22278 , 2024 2024 Citations: 63
Understanding machine learning & its application in obesity estimation by explainable AI MK Mahadi, R Rahad, A Noman, S Ishrat, F Faisal 2024 International Conference on Inventive Computation Technologies (ICICT … , 2024 2024 Citations: 17
Implementation of personal safety equipment tracking & detection by deepsort & yolov8 MK Mahadi, R Rahad, MS Mobassir, A Rahman, A Shafiullah, MM Nishat 2024 International Conference on Inventive Computation Technologies (ICICT … , 2024 2024 Citations: 6
Attention-based models for multivariate time series forecasting: Multi-step solar irradiation prediction S Sakib, MK Mahadi, SR Abir, AM Moon, A Shafiullah, S Ali, F Faisal, ... Heliyon 10 (6) , 2024 2024 Citations: 37
Fuel classification and adulteration detection using a highly sensitive plasmonic sensor R Rahad, AKM Rakib, MK Mahadi, MO Faruque Sensing and Bio-Sensing Research 40, 100560 , 2023 2023 Citations: 85
Machine learning assisted decision support system for prediction of prostrate cancer MK Mahadi, SR Abir, AM Moon, M Adnan, MANI Khan, MM Nishat, ... 2023 20th International Conference on Electrical Engineering/Electronics … , 2023 2023 Citations: 16
Employment of ensemble machine learning methods for human activity recognition T Hasan, MF Bin Karim, MK Mahadi, MM Nishat, F Faisal Journal of Healthcare Engineering 2022 (1), 6963891 , 2022 2022 Citations: 27
MOST CITED SCHOLAR PUBLICATIONS
A polarization independent highly sensitive metasurface-based biosensor for lab-on-chip applications R Rahad, MA Haque, MK Mahadi, MO Faruque, SMTS Afrid, ASM Mohsin, ... Measurement 231, 114652 , 2024 2024 Citations: 98
Fuel classification and adulteration detection using a highly sensitive plasmonic sensor R Rahad, AKM Rakib, MK Mahadi, MO Faruque Sensing and Bio-Sensing Research 40, 100560 , 2023 2023 Citations: 85
An alternative plasmonic material-based CMOS-compatible temperature sensor R Rahad, MM Sobhani, MJH Emon, SMTS Afrid, MK Mahadi, ASM Mohsin, ... Optics Communications 569, 130749 , 2024 2024 Citations: 63
Highly sensitive optically tunable transition metal nitride-based plasmonic pressure sensor with CMOS-compatibility at compact subwavelength dimensions R Rahad, MA Haque, MK Mahadi, ASM Mohsin, MO Faruque, SMTS Afrid, ... IEEE sensors journal 24 (14), 22271-22278 , 2024 2024 Citations: 63
Attention-based models for multivariate time series forecasting: Multi-step solar irradiation prediction S Sakib, MK Mahadi, SR Abir, AM Moon, A Shafiullah, S Ali, F Faisal, ... Heliyon 10 (6) , 2024 2024 Citations: 37
Employment of ensemble machine learning methods for human activity recognition T Hasan, MF Bin Karim, MK Mahadi, MM Nishat, F Faisal Journal of Healthcare Engineering 2022 (1), 6963891 , 2022 2022 Citations: 27
Gated recurrent unit (GRU)-based deep learning method for spectrum estimation and inverse modeling in plasmonic devices MK Mahadi, R Rahad, MA Haque, MM Nishat Applied Physics A 130 (11), 784 , 2024 2024 Citations: 21
Understanding machine learning & its application in obesity estimation by explainable AI MK Mahadi, R Rahad, A Noman, S Ishrat, F Faisal 2024 International Conference on Inventive Computation Technologies (ICICT … , 2024 2024 Citations: 17
Machine learning assisted decision support system for prediction of prostrate cancer MK Mahadi, SR Abir, AM Moon, M Adnan, MANI Khan, MM Nishat, ... 2023 20th International Conference on Electrical Engineering/Electronics … , 2023 2023 Citations: 16
RoadSense: a framework for road condition monitoring using sensors and machine learning IA Jahan, AS Huq, MK Mahadi, IA Jamil, MZ Shahriar IEEE Transactions on Intelligent Vehicles , 2024 2024 Citations: 14
Implementation of personal safety equipment tracking & detection by deepsort & yolov8 MK Mahadi, R Rahad, MS Mobassir, A Rahman, A Shafiullah, MM Nishat 2024 International Conference on Inventive Computation Technologies (ICICT … , 2024 2024 Citations: 6
A phishing detection approach for empowering cybersecurity with explainable ai and selectkbest MK Mahadi, R Rahad, MZ Shahriar, R Debnath, FN Rahman, AMT Haider, ... 2024 27th International Conference on Computer and Information Technology … , 2024 2024 Citations: 1
Explainable Transfer Learning for Precise Alzheimer’s Disease Prediction from MRI Data MK Mahadi, J Uddin, MZ Shahriar, R Rahad, AMT Haider, RH Ratul, ... 2024 International Conference on Innovations in Science, Engineering and … , 2024 2024 Citations: 1
BSAF-Based Binary Offloading for Edge AI: Energy-Efficient Deep Learning and Spatial Computing in Mixed Reality MK Mahadi, DM Doe, X Li, Z Han, L Qian 2026 International Conference on Computing, Networking and Communications … , 2026 2026
Dual-mode CMOS-compatible optically tunable plasmonic sensor with symmetrical cavity for simultaneous and independent monitoring of pressure and temperature deviations from … R Rahad, JD Joy, MS Rahman, MJH Emon, R Rahad, MK Mahadi Surfaces and Interfaces, 108325 , 2025 2025
Energy-Efficient Task Offloading Frameworks For Mixed Reality And Extended Reality In Edge AI Environments MK Mahadi Prairie View A&M University , 2025 2025
