My research develops and applies machine learning for cybersecurity, digital forensics, hardware and embedded systems, and health care, with an emphasis on models whose decisions can be explained and trusted. A second line of work, begun in my doctoral research at the University of Toledo on computational models inspired by the adaptive immune system, develops nature-inspired algorithms for training and designing neural networks.
At the Higher Colleges of Technology, my research centers on artificial intelligence (AI) for cybersecurity: machine learning for cyber threat detection, malware detection and analysis, digital forensics, hardware security, critical infrastructure security and anomaly detection. Related interests are trustworthy and explainable AI, secure embedded and intelligent systems, generative AI and digital twins.
I build datasets and detection models for machine-generated text, misinformation and clickbait, and natural language processing (NLP) methods that help investigators narrow a pool of suspects using text evidence. With students and visiting officer-cadets I have also explored machine learning for hardware Trojan detection, supervisory control and data acquisition (SCADA) network security and email spam detection.
Representative work: AI-generated versus human text (IEEE Transactions on Artificial Intelligence, 2025); psycholinguistic NLP for forensic text analysis (Frontiers in Artificial Intelligence, 2025); persons of interest in digital forensics (Future Internet, 2024); misinformation detection (Electronics, 2023).
I develop explainable models for clinical prediction and medical imaging, including Kolmogorov-Arnold networks for cardiovascular risk tiers, explainable prediction of diabetic distress, an uncertainty-aware ensemble for detecting intracranial hemorrhage on computed tomography (CT) scans, and the MultiCaRe-Thorax chest X-ray benchmark. In National Institutes of Health (NIH) AIM-AHEAD projects led by Norwich University with the OCHIN health center network, I contributed machine learning expertise to research on identifying exposure to firearm violence in electronic health records, including the natural language processing studies listed below.
Representative work: MultiCaRe-Thorax (Journal of Imaging, 2026); X-HEM (Computers, 2026); Kolmogorov-Arnold networks for carotid intima-media thickness (Scientific Reports, 2025); a text classifier protocol for firearm violence exposure in clinical notes (JMIR Research Protocols, 2025); firearm violence exposure in health records (JAMIA Open, 2024).
I apply clonal selection and swarm intelligence to train multilayer perceptrons, optimize the topologies of long short-term memory (LSTM) networks, design convolutional neural networks (CNNs) automatically and tune extreme learning machines for medical diagnosis.
Representative work: immunocomputing for LSTM topologies (IEEE Access, 2021); multilayer perceptron training with nature-inspired computing (IEEE Access, 2022); automated CNN architecture design (Mathematics, 2023); hybrid extreme learning machine framework (Expert Systems, 2024).
I have implemented neural network training, particle swarm optimization and ant colony optimization on field-programmable gate arrays (FPGAs) to accelerate learning and search.
Representative work: backpropagation on FPGA (IEEE National Aerospace and Electronics Conference, 2018); particle swarm optimization on FPGA (International Journal of Advanced Computer Science and Applications, 2019); parallel ant colony optimization for the traveling salesman problem (International Journal of Computer Applications in Technology, 2022).
With colleagues at Norwich University and the Virginia Military Institute, I study how natural language processing and machine learning can analyze student reflections and surveys in engineering courses.
Representative work: student reflections on a remote group project (2025) and attitudinal surveys on generative AI tools (2026), both at the American Society for Engineering Education (ASEE) Annual Conference.
My co-authors are at universities and research organizations in Jordan, India, Australia, Sweden, Kuwait and the United States, including the OCHIN health center network. I welcome new collaborations, particularly on AI for cybersecurity and on trustworthy AI in health care.
CiteSound is a free web tool I built that checks each entry of a reference list against Crossref, OpenAlex, Semantic Scholar, arXiv and PubMed. It reports whether each cited work exists and whether its details match the published record, and it flags fabricated references, real DOIs attached to the wrong work, and drifted details such as the year, authors, volume or pages. Open CiteSound