The rapid growth of cloud computing and the Internet of Things (IoT) has increased the exposure of IoT devices to cyber-attacks due to their resource limitations and lack of standardized security protocols. This paper presents a robust anomaly detection framework for IoT networks using two unsupervised machine learning models: Isolation Forest (IF) and One-Class Support Vector Machine (OCSVM). Leveraging the TON_IoT dataset, we conduct a comparative evaluation of IF, OCSVM, and a lightweight fusion approach called Combined Scoring Anomaly Detection (CSAD). Results show that OCSVM achieves superior precision, recall, and accuracy compared to both IF and CSAD. To ensure reliability, we apply Random Forest-based feature importance analysis, fivefold cross-validation and hyperparameter tuning. Model resilience is further examined under adversarial label-flip poisoning attacks and interpretability is enhanced through Local Interpretable Model-Agnostic Explanations (LIME). The findings demonstrate that lightweight unsupervised algorithms can provide effective, low-resource anomaly detection for modern IoT environments.
Rapid growth in technological criminal activities has drawn worldwide attention to cyber forensics. The objective of a cyber forensics is to provide situation awareness in terms of identification and preservation of digital evidence, extraction of information, and analysis of extracted information to facilitate time-critical decision making. However, Cyber Forensic Investigations (CFIs) still lack significant structure to provide reliable insight into major cyberattack patterns. Data from the Global Cyber Security Index (GCI) show that Pakistan lags far behind in technological and organizational initiatives, posing a threat to its national security. This article focuses on state coordinated CFI infrastructure to mitigate the occurrence of cybercrime challenges. Moreover, a conceptual model is established to address the infrastructure of CFI using policy approach, legal, technical, organizational, capacity building, and cooperative venture. This conceptual model is structured in pillars to simplify the CFI infrastructure. The goal of the study is to provide an empirical foundation for policymakers to develop a comprehensive framework for CFI in the country, and to offer insights and recommendations for improving the infrastructure to better combat cybercrime in Pakistan.
Accurately predicting student performance is essential for improving educational outcomes and guiding targeted interventions. This study applies eight advanced machine learning models-Decision Trees, Random Forest, Lasso, K-Nearest Neighbors, XGBoost, CatBoost, AdaBoost, and Gradient Boosting to analyze student performance based on demographic and academic features. Among these, CatBoost achieved the highest accuracy (87.46%) and less error rates, outperforming Gradient Boosting (87.28%) and Decision Trees (82.42%). Model evaluation was conducted using Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE), demonstrating the robustness of the proposed approach. The results highlight the effectiveness of data-driven methods in early identification of at-risk students, enabling educators to implement personalized learning strategies. This study underscores the transformative potential of machine learning in education, paving the way for more adaptive and student-centered learning environments.
In this article, we consider the problem of resilient base containment control for fractional-order multi-agent systems (FOMASs) with mixed time delays using a reliable and simple approach, where the communication topology among followers is a weighted digraph. A disturbance term is introduced into the delayed and non-delayed controller part to make it more practical. Our method involves proposing algebraic criteria by utilizing non-delayed and delayed protocols, applying the Razumikhin technique and graph theory respectively. The presented method can well overcome the difficulty resulting from fractional calculus, time delays and fractional derivatives. To demonstrate the validity and effectiveness of our findings, we provide an example at the end of our study.
Machine learning has become an essential component across various domains, including the education sector. Accurately predicting students’ academic performance plays a critical role for teachers and school administrators—not only in enhancing the quality of education but also in influencing educational outcomes. In this study, we propose an innovative system capable of predicting student achievement with high accuracy. We analyze a data set containing student-related features such as gender, race/ethnicity, parental education level, participation in breakfast programs, test preparation, participation in courses, computer science scores, and literacy skills. A supervised machine learning approach, specifically the random forest algorithm, was employed to build the prediction model. Redundant features were eliminated to reduce complexity and computational cost. Additionally, a comparative analysis was conducted to demonstrate the effectiveness of the proposed method. The findings of this research have the potential to transform educational decision-making and support a more data-driven and effective strategy for improving academic outcomes.