ABSTRACT In recent times, user privacy on social media networks is considered a serious concern, thus necessitating an efficient machine learning model for detecting anomalies. Many traditional approaches fall short in handling the increasing complexity of online threats, and this gap creates the need to integrate privacy‐aware access control models with machine learning systems. Therefore, this paper proposes a novel multimedia security and privacy analysis framework by utilizing a Machine Learning model for fog edge computing networks. In contrast to traditional Machine Learning‐based federated privacy models that consider the anomaly detection and privacy protection as loosely coupled components, the proposed framework creates a tightly integrated and interdependent security‐privacy system. In this system, threat scores govern privacy enforcement at the fog edge level. The Reinforced Gaussian Fuzzy Quantile Regression model performs the multimedia security analysis by integrating the Gaussian probability estimation and fuzzy rule‐based threat prediction. In addition, the Trust‐based Federated Markov Probability Differential Privacy model is developed by the integration of the Federated Learning, Markov Modeling and Differential privacy for privacy analysis. To validate the proposed model, three benchmark datasets are employed and the results highlight the superiority of a model with the accuracy of 98.07% and network efficiency of 98.17%. Overall, these findings confirm that the model provides a promising solution for securing multimedia data.
ElShafie et al. (Tue,) studied this question.