Federated Learning for Cybersecurity: A Promising Frontier

Abstract

In today’s world, Artificial Intelligence and Machine Learning are transforming many industries, but they rely on huge amounts of data leading to privacy issues. A better alternative is Federated learning. Federated Learning involves training the model from multiple sources using the decentralization technique, meaning each device trains the model on its local data thereby reducing the strain on the single server. This is useful in cases where the data is too large to be sent and maintained on a central server or in handling privacy and security concerns. Cybersecurity is a prime domain where vulnerable attacks and breaches can be prevented using this technique. Other domains like finance, healthcare, IoT etc. have transformed the idea of data-driven decisions. This paper gives an insight into the concept of Federated learning and why it is a better choice. It includes the various algorithms that can be implemented in multiple applications depicted through a comparative analysis. The choice of algorithm depends upon its efficiency and the desired cybersecurity application.