Emmanuel Ajayi Olajubu

The cyber infrastructure of Industrial Control Systems (ICSs) that monitor power grids, water treatment facilities, pipelines, and manufacturing lines has become a very com-plex cyber-physical systems, which now face a growing range of cyber threats, such as Stuxnet, Industroyer, TRITON, and Colonial Pipeline incident. In the era of new and sophisticated attacks, such as zero-day exploits, multi-stage intrusions and adversaries taking advantage of the unique ‘availability first' constraints of operational technology (OT), traditional signatures and rule-based defenses are proving less effective. This re-view is systematic, analyzing the use of artificial intelligence (AI) methodologies in ICS cybersecurity from the year 2018 to 2024. According to the guidelines of the PRISMA 2020, from the five scholarly databases, 1578 records were retrieved and 147 studies were included in the qualitative synthesis, and 102 studies were included in the quantitative comparison. The provided literature is categorized in a methodological taxonomy from classical machine learning to deep learning; from graph neural networks (GNNs) to reinforcement learning (RL); from autonomous response to intrusion detection, at-tack-graph analysis, malware analysis and vulnerability prioritization. Overall, the synthesis suggests that graph-based and hybrid methods yield the best detection accuracy (reported from 92% to 99.5% with less than 3% false-positive rates), whereas classical methods still seem to be the most suitable for resource-constrained applications. There are ongoing deficiencies in standardized benchmarking, adversarial robustness, safety–security co-engineering, and real-world validation. A research roadmap is advocated for guiding future research, which focuses on safe RL, federated learning, explainable GNNs, and digital-twin-based security assessment.

The increasing integration of Industrial Control Systems (ICS) with Information Technology (IT) networks has significantly expanded the cybersecurity attack surface of critical infrastructure. Conventional cybersecurity risk management approaches, which primarily rely on periodic assessment and static auditing mechanisms, are often inadequate for addressing dynamic and evolving cyber threats in modern ICS environments. This study proposes a Dynamic Graph-Theoretic Cybersecurity Risk Management Model (ICS-CRMM) that integrates Graph Theory, Graph Neural Networks (GNNs), and Reinforcement Learning (RL) to enable adaptive threat detection and automated mitigation in real time. The proposed framework represents ICS environments as dynamic attack graphs, where nodes (e.g., PLCs, HMIs, and sensors) and communication links are weighted using vulnerability severity, exploitability, and operational impact factors. To quantify cybersecurity risk, the framework introduces a unified Graph Risk Score (GRS) derived from three complementary components: Node Risk Score (NRS), Edge Risk Score (ERS), and Path Risk Score (PRS). The resulting risk values are normalized using sigmoid-based probabilistic scoring to support continuous and topology-aware risk assessment. The framework was evaluated using an OpenPLC-based water tank cyber range under multiple attack scenarios, including Command Hijack, Man-in-the-Middle (MitM), Data Exfiltration, False Data Injection, and Distributed Denial-of-Service (DDoS) attacks. Experimental results demonstrated high detection performance, achieving 99.98% classification accuracy, a False Alarm Rate (FAR) of 0.01%, a Missed Attack Rate (MAR) of 0.02%, and sub-second mitigation response times within the controlled experimental environment. The integrated RL agent further enabled autonomous and Safety Integrity Level (SIL)-aware mitigation while maintaining operational continuity. The findings demonstrate the potential of graph-based and topology-aware artificial intelligence approaches for enhancing adaptive cybersecurity risk management in industrial environments. The proposed framework provides a practically deployable and operationally interpretable solution for improving the resilience and security of critical infrastructure systems against evolving cyber threats.

In today’s digital age, threats to life have become alarmingly common, infiltrating everything from social media platforms to personal communications. About 11.3% of people experienced suicidal ideation over the past year, 5.5% of adults in the United States reported serious suicidal thoughts in the past year, and 10-14% of children and adolescents experienced cyberbullying globally. This study developed machine learning models to detect life-threatening statements across different model categories by employing hyperparameter tuning and Adam optimization. Six models were built using RF, SVM, FFNN, LSTM, pre-trained BERT, and T5, and evaluated using benefit and cost measures and confusion matrices. The outcome indicates that the six models performed well, but the FFNN model outperformed all other models by achieving an accuracy of 98.86%, precision of 98.86%, MCC of 97.72%, Recall of 98.86%, MAE of 0.0113, and RMSE of 0.1067. It also achieved the highest number of accurately classified instances and the lowest number of misclassified instances, which demonstrates that the FFNN model exhibits robustness and precision, making it highly reliable for detecting life-threatening statements compared to other models. Having achieved a very high accuracy, the method still lacks an in-depth understanding of contextual information. Therefore, further studies may focus on incorporating contextual information for more accurate classification of life-threatening statements and integrating the developed model into a real-time detection system.

Handwritten digit classification represents a foundational task in computer vision and has been widely adopted in applications ranging from Optical Character Recognition (OCR) to biometric authentication. Despite the availability of large benchmark datasets, the development of models that achieve both high accuracy and computational efficiency remains a central challenge. In this study, the performance of three representative machine learning paradigms—Chi-Squared Automatic Interaction Detection (CHAID), Generative Adversarial Networks (GANs), and Feedforward Deep Neural Networks (FFDNNs)—was systematically evaluated on the Modified National Institute of Standards and Technology (MNIST) dataset. The assessment was conducted with a focus on classification accuracy, computational efficiency, and interpretability. Experimental results demonstrated that deep learning approaches substantially outperformed traditional Decision Tree (DT) methods. GANs and FFDNNs achieved classification accuracies of approximately 97%, indicating strong robustness and generalization capability for handwritten digit recognition tasks. In contrast, CHAID achieved only 29.61% accuracy, highlighting the limited suitability of DT models for high-dimensional image data. It was further observed that, despite the computational demand of adversarial training, GANs required less time per epoch than FFDNNs when executed on modern GPU architectures, thereby underscoring their potential scalability. These findings reinforce the importance of model selection in practical deployment, particularly where accuracy, computational efficiency, and interpretability must be jointly considered. The study contributes to the ongoing discourse on the role of artificial intelligence (AI) in pattern recognition by providing a comparative analysis of classical machine learning and deep learning approaches, thereby offering guidance for the development of reliable and efficient digit recognition systems suitable for real-world applications.

A contrived dataset of substation automation for cybersecurity research in the smart grid networks based on IEC61850

The widespread expansion of the internet has fueled a global surge in the utilization of various online transactions, with a significant portion of such services operating on mobile web platforms. Simultaneously, the deployment of innovative Mobile Ad Hoc Network (MANET) technologies and mobile applications has grown as solutions for diverse tasks. Unfortunately, this progress has attracted the attention of hackers, who continually devise new strategies to exploit the vulnerabilities inherent in mobile networks. This study aims to address the escalating challenges posed by cyber threats in the era of widespread internet expansion, particularly focusing on securing mobile web platforms against sophisticated attacks such as Structured Query Language Injection (SQLi) for web-based database solutions and Denial of Service (DoS/DDoS) for various applications. In response to the identified vulnerabilities, this paper proposes an HTTP regeneration (HReg) model that not only detects various cyber-attacks but also ensures the uninterrupted provision of critical services during such incidents. The model introduces an innovative regeneration algorithm capable of scanning both the connection channel and web application to detect attacks, creating survivable connections within the underlying TCP engine to replace compromised ones during an ongoing attack. In simulated environments using OMNeT++, where the server is subjected to attacks, the experimental results demonstrate the efficacy of the model. The response and performance metrics, including throughput (73%), delivery ratio (68.8%), delay (3s), and network load, showcase the model's ability to detect and neutralize attacks. A comparison with state-of-the-art approaches highlights the superior performance of the regeneration model, attributed to its additional survivability layer. While the regeneration model proves robust in simulated environments, its real-world application may encounter limitations. Future research should explore these limitations to enhance the practical applicability of the proposed model. The proposed HReg model's resilient performance under attack conditions ensures the survivability of web-based applications. This innovative approach offers practical implications for securing mobile web platforms, providing continuous delivery of critical services even in the face of persistent and evolving cyber threats. This research addresses a significant gap in existing efforts by not only focusing on attack detection but also emphasizing the development of a survivable TCP connection for HTTP servers during attacks. The introduced regeneration model stands out for its unique ability to maintain service continuity, showcasing originality in the approach to cybersecurity in the context of web-based applications.

Mobile wireless networks enable the connection of devices to a network with minimal or no infrastructure. This comes with the advantages of ease and cost-effectiveness, thus largely popularizing the network. Notwithstanding these merits, the open physical media, infrastructural-less attributes, and pervasive deployment of wireless networks make the channel of communication (media access) vulnerable to attacks such as traffic analysis, monitoring, and jamming. This study designed a virtual local area network (VLAN) model to circumvent virtual jamming attacks and other intrusions at the Media Access Control (MAC) layer of the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) protocol. A Torpor VLAN (TVLAN) Data Frame Encapsulation and the algorithm for T-VLAN security in CSMA/CA were formulated and presented. A simulation experiment was conducted on the model using OMNeT++ software. The performance metrics used to evaluate the model were packet delivery ratio, network throughput, end-to-end channel delay, and channel load. The simulation results show that the TVLAN defence mechanism did not increase the channel load arbitrarily during TVLAN defence. similarly, the system throughput was shown to be 82% during TVLAN defence. Nevertheless, the network delay of the system during TVLAN defence was significantly high but the channel load was 297 when the TVLAN security mechanism was launched. These results demonstrate the model’s ability to provide a survivability mechanism for critical systems when under attack and add a security layer to the CSMA/CA protocol in wireless networks. Such a remarkable performance is required of a CSMA/CA infrastructure for improving the cybersecurity posture of a wireless network.

Diabetes is one of the major health issues that affect more than 10.5 percent of the adult population across the globe. This study applied deep learning techniques of deep belief network (DBN), long short-term memory (LSTM) and recurrent neural network (RNN) with Bayesian optimization on a diabetes dataset to forecast patients with diabetes. A splitting ratio of 80:20 was used for model performance evaluation. DBN model had the lowest mean absolute error in comparison to the other two models with 95.79% accuracy, 0.0331 mean absolute error, 0.0709 mean squared error, 0.1204 loss function, 0.9458 precision, 0.1819 RMSE, and 0.5307 recall. The results from this study validate that the DBN model can be used on a larger dataset to reduce variance and model overfitting, thereby achieving a better accuracy score.