A SYSTEMATIC REVIEW OF PRESERVING PRIVACY IN FEDERATED LEARNING: A REFLECTIVE REPORT - A COMPREHENSIVE ANALYSIS

Abstract

 The primary purpose of this research is to systematically evaluate the technology and approaches that presently define privacy preservation in Federated Learning's situation. Federated learning, an alternative to centralized machine learning, has compelling applications where privacy-sensitive data is to be combined for collaborative model training in a distributed manner [1]. While privacy preservation is a pressing issue in federated learning systems, the current solutions should be carefully evaluated to ensure that the proper methodology is used. This reflective report summarizes the recent developments in privacy-preserving aspects of federated Learning, highlighting major critiques, hurdles, and recommendations. Decentralized Learning is an emerging approach to machine learning where models are distributed among devices to be trained while keeping data confidential [1]. The subsequent section builds upon the thought-awakening part, which addresses the issue of privacy in federated Learning. A literature review is done systematically to evaluate the techniques and methods for maintaining privacy in the context of federated Learning. Important aspects include differential privacy, secure aggregation protocols, encryption technologies, and Federation learning design. The manuscript delves into the hurdles and openings regarding protecting privacy in federated Learning and directs future work on the subject [1,2]. This report combines the available knowledge and pinpoints the possible gaps, thus contributing to more profound knowledge about the privacy issues of federated Learning as presented in the recommendations.