Explore federated learning, a distributed machine learning approach that preserves data privacy, in this 44-minute conference talk from Strange Loop. Learn how this technique enables collaboration on ML models without sharing sensitive data directly. Discover the federated averaging algorithm, real-world challenges, and ongoing research to enhance security, reduce communication costs, and strengthen privacy guarantees. Gain insights into practical applications, potential pitfalls, and strategies for implementing federated learning across various domains, from embedded devices to legal entities. Understand when and how to leverage this technology to balance the benefits of machine learning with crucial privacy concerns.