Enabling Privacy-Centric AI Models for Secure Data Analysis in Federated Machine Learning Frameworks

Abstract

With the increasing demand for privacy preservation in the data-driven world, federated learning (FL) has emerged as a promising solution that allows machine learning models to be trained across decentralized devices without the need to share sensitive data. However, ensuring privacy while leveraging AI models in federated frameworks remains a significant challenge. This paper investigates the integration of privacy-centric artificial intelligence (AI) models in federated machine learning systems to enhance data security during model training and data analysis. By focusing on techniques such as differential privacy, secure multi-party computation (SMPC), and homomorphic encryption, the paper explores the potential of these privacy-preserving mechanisms in safeguarding sensitive information without compromising model performance. The study evaluates the advantages and limitations of these techniques, and presents a framework for integrating privacy-focused AI models within federated learning systems to enable secure and efficient data analysis. The paper also highlights key challenges in achieving scalability, model accuracy, and computational efficiency in privacy-centric federated learning setups.