An Efficient Reliable Federated Learning Technology

Many organizations in this competitive world use data analysis to provide useful insights for decision making and problem solving tasks that exist in real - time scenarios. To preserve the information provided, technological improvements are needed to handle data privacy in the system. In real-world circumstances, large amounts of data will be collected from a variety of sources and need to be analyzed and protected from unauthorized access. As a result, machine learning models must be deployed across all sites without the need to upload data, as this could compromise privacy. Federated learning helps to provide this privacy for the data. In this paper, we have analyzed two methods of building machine learning models - the centralized or conventional method of building machine learning models and the federated approach of building models. The performance metrics in terms of accuracy has been examined. The results are used to highlight the improved accuracy of federated approach compared to the centralized approach without compromising the privacy of the data.