Design of an efficient dynamic context-based privacy policy deployment model via dual bioinspired Q learning optimisations

Abstract

A novel context-based privacy policy deployment model enhanced with bioinspired Q-learning optimisations is presented. The model addresses the challenge of maintaining privacy while ensuring data integrity and usability in various settings. Leveraging datasets including Adult (Census Income), Yelp, UC Irvine Machine Learning, and Movie Lens, the authors evaluate the model's performance against state-of-the-art techniques, such as GEF AL, Deep Forest, and Robust Continual Learning. The approach employs Firefly Optimiser (FFO) and Ant Lion Optimiser (ALO) algorithms to dynamically adjust privacy parameters and handle large datasets efficiently. Additionally, Q-learning enables intelligent decision-making and rapid adaptation to changing data and network conditions and scenarios. Evaluation results demonstrate that the model consistently outperforms reference techniques across multiple metrics, including privacy levels, scalability, fidelity, and sensitivity management. By reducing reputational harm, minimising delays, and enhancing network quality, the model offers robust privacy protection without sacrificing data utility. Overall, a dynamic context-based privacy policy deployment approach, enhanced with bioinspired Q-learning optimisations, presents a significant advancement in privacy preservation methods. The combination of ALO, FFO, and Q-learning techniques offers a practical solution to evolving data privacy challenges and enhances flexibility in various use case scenarios.