A comparative study of machine learning, deep learning algorithms, and explainable AI techniques for diabetes prediction Muhammad Imad, Muhammad Shakeel, Hamail Raza Zaidi, Zabih Ullah Khan Utilizing AI of Medical Things for Healthcare Security and Sustainability, 2025 Diabetes prediction remains a crucial area of research due to its profound impact on global health. Diabetes, a chronic metabolic disorder, affects millions of people worldwide and poses significant challenges to healthcare systems. Early prediction and diagnosis are essential to managing the disease effectively, preventing complications, and improving the quality of life for patients. Recent advancements in artificial intelligence (AI) have paved the way for powerful tools in diabetes prediction, particularly through machine learning and deep learning algorithms. These methods offer promising solutions for enhancing early diagnosis and personalized care.
Lightweight Wildlife Image Classification Using EfficientNetV2-S with Confidence-Aware Prediction Ravikanth Manchana, Ali Gohar, Muhammad Imad, Raja Hashim Ali 2025 27th International Multitopic Conference Inmic 2025, 2025 Automated classification of wildlife images is essential for conservation, biodiversity monitoring, and ecological research, where large datasets are collected from camera traps and drones. Existing solutions often rely on computationally heavy pretrained models and force predictions even for unfamiliar inputs, limiting real-world usability. This study proposes a lightweight and practical wildlife species classification pipeline using EfficientNetV2-S, trained entirely from scratch in a resource-constrained Kaggle environment. The dataset comprised 5,175 images spanning 18 species, each provided at multiple resolutions (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$224 \times 224,\ 300 \times 300$</tex>, and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$512 \times 512$</tex>). Preprocessing included normalization and basic augmentation, while training employed checkpointing to retain the best-performing model. To improve reliability, a confidence thresholding mechanism was integrated during inference, which allowed the system to reject uncertain predictions and return “Unknown” for out-ofdistribution inputs. Experimental results achieved 5 2% validation accuracy. Comparative analysis with MobileNetV2 and ResNet18 showed that EfficientNetV2-S achieved superior balance between accuracy and computational efficiency. Moreover, thresholding effectively reduced false positives in unfamiliar species tests. This study demonstrates a deployable, safe, and efficient deep learning approach for wildlife classification, with direct applicability in ecological monitoring, anti-poaching systems, and conservation field deployments.
Deep Learning-Based Tomato Detection from Images for Automated Farming Fanglei Zhou, Ali Raza, Muhammad Imad, Raja Hashim Ali 2025 27th International Multitopic Conference Inmic 2025, 2025 Object detection is a critical area of computer vision. It has several applications where it supports many applications in agriculture, food processing, and automation. An important application is the accurate detection of tomatoes, which is essential for improving efficiency as well as in reducing reliance on manual labor in agricultural workflows. While deep learning models have advanced general object detection, there remains limited research focused on tomato identification using modern frameworks. Moreover, researchers have explored enhanced YOLO versions that integrate attention modules to improve fruit detection under occlusion and varying illumination. This study aims to address this gap by developing a detection pipeline based on the YOLOv8 Nano architecture. The dataset was collected from publicly available sources and converted from Pascal VOC annotations to YOLO format to ensure compatibility with the model. The YOLOv8 Nano model was trained for five epochs with a batch size of sixteen and an input resolution of six hundred forty pixels, utilizing GPU acceleration for improved performance. Evaluation results showed a mean average precision of approximately eightynine and high detection confidence across the validation images. These findings confirm that lightweight deep learning models can achieve strong detection capabilities in specialized agricultural tasks. The study contributes a practical approach to automating tomato detection workflows and offers a foundation for further research in efficient agricultural applications.
End-to-End Detection and Generative Modelling of Exoplanet Transits Using Recurrent Neural Networks Ilnaza Saifutdinova, Aiman Darakhshan, Muhammad Imad, Raja Hashim Ali 2025 International Conference on Frontiers of Information Technology Fit 2025, 2025 The detection and interpretation of exoplanet transit signals remain central challenges in astrophysics, particularly given the scale and noise of modern light curve datasets. This paper presents a unified deep learning framework that combines supervised detection and unsupervised generative modeling of stellar transits. The proposed architecture integrates a Long Short-Term Memory (LSTM) classifier with a symmetric LSTM autoencoder, enabling simultaneous classification, reconstruction, and synthesis of transit signals. Evaluated on the Kepler light curve dataset, the classifier achieved strong performance (F1-score = 0.94, AUC = 0.98), while the autoencoder preserved essential transit features with low reconstruction error (MSE = 0.00279). Latent space analysis revealed compact and structured embeddings that enabled decoding into coherent synthetic light curves, demonstrating the generative potential of the model. Unlike conventional approaches that separate detection from signal modeling, this dual-module system bridges both tasks within a single pipeline, enhancing interpretability and offering tools for simulation and anomaly detection. These results highlight the effectiveness of recurrent architectures in astrophysical time-series analysis and support their application to future exoplanet discovery missions, where detection accuracy, reconstruction fidelity, and interpretability are equally critical.
A Comparative Evaluation of Search and Metaheuristic Algorithms for the N-Queens Problem: Scalability, Efficiency, and Success Rates Eslam Mahmoud Mohamed Mahmoud Aly, Muhammad Maaz Hamid, Muhammad Imad, Raja Hashim Ali 2025 27th International Multitopic Conference Inmic 2025, 2025 The N-Queens problem is a well-known test in artificial intelligence and optimization, it involves the arranging N queens on NxN chessboard so that no two queens attack each other. This study offer a replicable comparison of four distinct algorithmic approaches: Depth-First Search (DFS), Simulated Annealing (SA), Hill Climbing (HC), Genetic Algorithm (GA). Each algorithm was implemented with the same unified experimental setup and tested across increasing board sizes (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$N=10,30$</tex>, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$50,100,200$</tex>) using the same identical initial states to maintain a fair comparison. the evaluation focused on a critical performance indicators such as execution time, number of moves, memory consumption, and success rate. The results show that while DFS performs reliably on smaller boards, it becomes impractical as N increases. HC with random restarts and SA both performed effectively on larger board sizes, SA demonstrating better time efficiency but needing more moves. Although GA achieved low memory consumption and fast iterations, it struggled to find valid solutions beyond <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$N=30$</tex>, even with increased number of generations. this work delivers a reproducible benchmarking framework that supports fair algorithmic evaluation and highlights the trade-offs between efficiency and scalability across different optimization methods. the insights gained from this study can inform future research in constraint satisfaction and hybrid metaheuristic strategies.
Solving the N-Queens Puzzle: A Performance Study of Classical and Metaheuristic Algorithms Liana Mikhailova, Syed Ghazi Abbas, Muhammad Imad, Raja Hashim Ali 2025 International Conference on Frontiers of Information Technology Fit 2025, 2025 The N-Queens problem is a foundational combinatorial optimization challenge used to test search and optimization algorithms. While basic in computer science, it has applications in scheduling, parallel processing, and constraint satisfaction. Exhaustive search provides solutions but struggles with scalability for larger boards. This paper investigates the potential of genetic algorithms (GAs) as a substitute method, demonstrating their strengths in flexibility and expandability, although, their stochastic nature remains a drawback. Research examining the brute force and GAs for various board sizes (N=10,50,100) demonstrated that exhaustive search is completely deterministic, but the complexity grows with factorial number. GAs get solutions close to the best possible solutions quicker for medium sizes, however, lack effects for large N due to the slow convergence. Results emphasize the trade-off between deterministic and probabilistic methods and provide insights into GA improvement tactics in complex situations.
A step toward the detection of alzheimer's disease using ensemble learning Artificial Intelligence for Intelligent Systems Fundamentals Challenges and Applications, 2024
Integrating Machine Learning and Deep Learning Approaches for Efficient Malware Detection in IoT-Based Smart Cities Journal of Computing and Biomedical Informatics, 2023
Cruising into the Future: Navigating the Challenges and Advancements in Autonomous Vehicle Technology Journal of Computing and Biomedical Informatics, 2023
Virtual Reality Based Interior Designing Using Amazon Web Services Raja Hashim Ali, Ali Zeeshan Ijaz, Muhammad Huzaifa Shah, Nisar Ali, Muhammad Imad, Said Nabi, Kiran Perveen, Javaria Tahir, Memoona Saleem 2023 International Conference on IT and Industrial Technologies Icit 2023, 2023
Robust and Reliable Liveness Detection Models for Facial Recognition Systems Haris Anjum, Usama Arshad, Raja Hashim Ali, Zain Ul Abideen, Muhammad Huzaifa Shah, Talha Ali Khan, Ali Zeeshan Ijaz, Abu Bakar Siddique, Muhammad Imad Proceedings 2023 International Conference on Frontiers of Information Technology Fit 2023, 2023
Task and Billing Automation System Mustafa Ali Mir, Ahmed Ali, Komal Ata, Muhammad Imad, Muhammad Naseem Icisct 2020 2nd International Conference on Information Science and Communication Technology, 2020
RECENT SCHOLAR PUBLICATIONS
An Impact of Sustainable Finance Initiatives on Financial Markets and Economic Stability M Hilal, Sumbal, M Imad Emerging Disruptive Technologies for Society 5.0 in Developing Countries … , 2025 2025 Citations: 1
Strategic Imperative of Data Analytics: Empowering Informed Business Decision-Making M Hilal, Sumbal, M Imad Emerging Disruptive Technologies for Society 5.0 in Developing Countries … , 2025 2025 Citations: 1
A Comparative Study of Machine Learning, Deep Learning Algorithms, and Explainable AI Techniques for Diabetes Prediction M Imad, M Shakeel, HR Zaidi, ZU Khan Utilizing AI of Medical Things for Healthcare Security and Sustainability … , 2025 2025
Integrating Traditional Machine Learning Approaches with Explainable Anomaly Detection for Multimodal Sensor Data M Imad, I Cleland, C Nugent, P McAllister International Conference on Ubiquitous Computing and Ambient Intelligence … , 2024 2024 Citations: 1
A comprehensive survey of machine learning applications in healthcare SH Bangash, M Imad Cognitive Machine Intelligence, 245-269 , 2024 2024 Citations: 1
Improving the accuracy of Anomaly Detection in Multimodal Sensors using 1D-CNN M Imad, I Cleland, P McAllister, C Nugent Proceedings of the 17th International Conference on PErvasive Technologies … , 2024 2024 Citations: 4
Enhanced ECG Anomaly Detection: Leveraging a fine-tuned CNN-AUTOENCODER P Imad, Muhammad, Cleland, Ian, Nugent, Chris, McAllister Northern Ireland Biomedical Engineering Society (NIBES) Symposium , 2024 2024
Cloud-enhanced machine learning for handwritten character recognition in dementia patients M Hasnain, V Gude, MO Edeh, F Masood, WU Khan, M Imad, NO Fidelia Driving Transformative Technology Trends With Cloud Computing, 328-341 , 2024 2024 Citations: 20
Integrating machine learning and deep learning approaches for efficient malware detection in IoT-Based smart cities SH Bangash, D Khan, A Ishtiaq, M Imad, M Tahir, W Ahmad, G Husnain, ... Journal of Computing & Biomedical Informatics 5 (02), 280-299 , 2023 2023 Citations: 8
Cruising into the Future: Navigating the Challenges and Advancements in Autonomous Vehicle Technology SH Bangash, G Husnain, A Nawaz, M Tahir, M Imad, ZU Khan, D Khan, ... Journal of Computing & Biomedical Informatics 5 (02), 114-135 , 2023 2023 Citations: 3
POX and RYU controller performance analysis on software defined network N Naim, M Imad, MA Hassan, MB Afzal, S Khan, AU Khan EAI Endorsed Transactions on Internet of Things 9 (3), e5 , 2023 2023 Citations: 16
Machine learning algorithms for classification of COVID-19 using chest X-ray images SH Badshah, M Imad, IU Khan, MA Hassan Advanced AI and internet of health things for combating pandemics, 85-96 , 2023 2023 Citations: 3
Comparative Analysis of Machine Learning Methods for Multi-Label Skin Cancer Classification M Imad, ZA Khan, SH Bangash, IU Khan, S Ahmad, A Ishtiaq 2023 9th International Conference on Information Technology Trends (ITT … , 2023 2023 Citations: 3
Machine Learning Solution for Orthopedics: A Comprehensive Review SHBN Muhammad Imad*, Muhammad Abul Hassan Machine Intelligence for Internet of Medical Things: Applications and Future … , 2023 2023
Smart Cane: Obstacle Recognition for Visually Impaired People Based on Convolutional Neural Network MI Adnan Hussain, Bilal Ahmad Machine Intelligence for Internet of Medical Things: Applications and Future … , 2023 2023 Citations: 2
Design a framework for IoT-Identification, Authentication and Anomaly detection using Deep Learning: A Review A Shoukat, MA Hassan, M Rizwan, M Imad, SH Ali, S Ullah EAI Endorsed Transactions on Smart Cities 7 (1), e2 , 2023 2023 Citations: 3
Investigation of Blockchain for COVID-19: A Systematic Review, Applications and Possible Challenges. SH Badshah, M Imad, MA Hassan, S Ullah, SH Ali EAI Endorsed Transactions on Smart Cities 7 (1) , 2023 2023 Citations: 1
A comparative analysis of intrusion detection in IoT network using machine learning M Imad, M Abul Hassan, S Hussain Bangash, Naimullah Big data analytics and computational intelligence for cybersecurity, 149-163 , 2022 2022 Citations: 27
New advancements in cybersecurity: A comprehensive survey MA Hassan, S Ali, M Imad, S Bibi Big Data Analytics and Computational Intelligence for Cybersecurity, 3-17 , 2022 2022 Citations: 21
Impact of routing techniques and mobility models on flying ad hoc networks MA Hassan, M Imad, T Hassan, F Ullah, S Ahmad Computational Intelligence for Unmanned Aerial Vehicles Communication … , 2022 2022 Citations: 10
MOST CITED SCHOLAR PUBLICATIONS
COVID-19 classification based on Chest X-Ray images using machine learning techniques M Imad, N Khan, F Ullah, MA Hassan, A Hussain Journal of Computer Science and Technology Studies 2 (2), 01-11 , 2020 2020 Citations: 49
Energy efficient hierarchical based fish eye state routing protocol for flying ad-hoc networks MA Hassan, SI Ullah, A Salam, AW Ullah, M Imad, F Ullah Indonesian Journal of Electrical Engineering and Computer Science 21 (1 … , 2021 2021 Citations: 33
COVID-19 lung image classification based on logistic regression and support vector machine Nazish, SI Ullah, A Salam, W Ullah, M Imad European, Asian, Middle Eastern, North African Conference on Management … , 2021 2021 Citations: 31
A comparative analysis of intrusion detection in IoT network using machine learning M Imad, M Abul Hassan, S Hussain Bangash, Naimullah Big data analytics and computational intelligence for cybersecurity, 149-163 , 2022 2022 Citations: 27
Multi-class classification for the identification of COVID-19 in X-ray images using customized efficient neural network A Hussain, M Imad, A Khan, B Ullah AI and IoT for Sustainable Development in Emerging Countries: Challenges and … , 2022 2022 Citations: 24
Pakistani currency recognition to assist blind person based on convolutional neural network M Imad, F Ullah, MA Hassan Journal of Computer Science and Technology Studies 2 (2), 12-19 , 2020 2020 Citations: 23
New advancements in cybersecurity: A comprehensive survey MA Hassan, S Ali, M Imad, S Bibi Big Data Analytics and Computational Intelligence for Cybersecurity, 3-17 , 2022 2022 Citations: 21
Diagnosing of dermoscopic images using machine learning approaches for melanoma detection A Salam, F Ullah, M Imad, MA Hassan 2020 IEEE 23rd International Multitopic Conference (INMIC), 1-5 , 2020 2020 Citations: 21
Cloud-enhanced machine learning for handwritten character recognition in dementia patients M Hasnain, V Gude, MO Edeh, F Masood, WU Khan, M Imad, NO Fidelia Driving Transformative Technology Trends With Cloud Computing, 328-341 , 2024 2024 Citations: 20
IoT based machine learning and deep learning platform for COVID-19 prevention and control: A systematic review M Imad, A Hussain, MA Hassan, Z Butt, NU Sahar Ai and IoT for Sustainable Development in Emerging Countries: Challenges and … , 2022 2022 Citations: 18
POX and RYU controller performance analysis on software defined network N Naim, M Imad, MA Hassan, MB Afzal, S Khan, AU Khan EAI Endorsed Transactions on Internet of Things 9 (3), e5 , 2023 2023 Citations: 16
Automatic detection of bullet in human body based on X-ray images using machine learning techniques M Imad, SI Ullah, A Salam, WU Khan, F Ullah, MA Hassan International Journal of Computer Science and Information Security (IJCSIS … , 2020 2020 Citations: 12
Navigation system for autonomous vehicle: A survey M Imad, MA Hassan, H Junaid, I Ahmad Journal of Computer Science and Technology Studies 2 (2), 20-35 , 2020 2020 Citations: 11
Impact of routing techniques and mobility models on flying ad hoc networks MA Hassan, M Imad, T Hassan, F Ullah, S Ahmad Computational Intelligence for Unmanned Aerial Vehicles Communication … , 2022 2022 Citations: 10
Integrating machine learning and deep learning approaches for efficient malware detection in IoT-Based smart cities SH Bangash, D Khan, A Ishtiaq, M Imad, M Tahir, W Ahmad, G Husnain, ... Journal of Computing & Biomedical Informatics 5 (02), 280-299 , 2023 2023 Citations: 8
Performance Analysis of POX and RYU Based on Dijkstra’s Algorithm for Software Defined Networking Naimullah, SI Ullah, AW Ullah, A Salam, M Imad, F Ullah European, Asian, Middle Eastern, North African Conference on Management … , 2021 2021 Citations: 7
A vision based road blocker detection and distance calculation for intelligent vehicles F Ullah, SI Ullah, A Salam, WU Khan, M Imad, MA Hassan International Journal of Computer Science and Information Security (IJCSIS … , 2020 2020 Citations: 5
Improving the accuracy of Anomaly Detection in Multimodal Sensors using 1D-CNN M Imad, I Cleland, P McAllister, C Nugent Proceedings of the 17th International Conference on PErvasive Technologies … , 2024 2024 Citations: 4
Cruising into the Future: Navigating the Challenges and Advancements in Autonomous Vehicle Technology SH Bangash, G Husnain, A Nawaz, M Tahir, M Imad, ZU Khan, D Khan, ... Journal of Computing & Biomedical Informatics 5 (02), 114-135 , 2023 2023 Citations: 3
Machine learning algorithms for classification of COVID-19 using chest X-ray images SH Badshah, M Imad, IU Khan, MA Hassan Advanced AI and internet of health things for combating pandemics, 85-96 , 2023 2023 Citations: 3
