A Machine Learning Framework for Detecting and Mitigation of Cyber Threats in IoT Environments
Kritika Murugandi Reddiar Seetharaman, Pranay Yadav · 2025
Artificial intelligence (AI) and machine learning are becoming more and more important in enhancing threat detection, response, and proactive defense as cybersecurity quickly changes. Two prominent solutions, leverage ML to solve Cybect Threat Intelligence (CTI) issues. A machine learningbased approach for identifying and reducing cyber threats in IoT settings is presented in this research, with an emphasis on the categorization of cyberattacks using the ResNet model. The framework is evaluated employing the CICDS 2019 dataset, and the ResNet model outperforms traditional machine learning models such as Feedforward Neural Network (FFNN), Support Vector Machine (SVM), and Gated Recurrent Unit (GRU), achieving an accuracy of 99.5% and an AUC of 93.3%. Although it achieved high accuracy, the discriminatory power of the ResNet model is less than optimum, as seen from the ROC plot; more work needs to be done. This paper has shown how DL can be applied effectively to IoT cyber threat detection and proposes research and development areas for enhancement, such as different data sets, tuning models of DL, and light models in the IoT environment.