Instruction: Provide a clear definition of Federated Learning and discuss at least two advantages it has over traditional centralized learning approaches.
Context: This question assesses the candidate's understanding of the fundamental concept of Federated Learning. It aims to evaluate their knowledge of how Federated Learning operates and why it is beneficial compared to centralized models, particularly in terms of privacy preservation and data security.
The way I'd explain it in an interview is this: Federated learning is a training approach where model updates are computed locally on client devices or institutions, and only model updates or gradients are sent back for aggregation instead of raw data. The core advantage is that data stays closer to where it was generated.
Compared with centralized learning, that can reduce data movement, improve privacy posture, and make collaboration possible across parties that cannot easily pool data. It is especially useful when legal, trust, or bandwidth constraints make central collection unattractive.
What matters in an interview is not only knowing the definition, but being able to connect it back to how it changes modeling, evaluation, or deployment decisions in practice.
A weak answer says federated learning means training on many devices, without explaining local training, central aggregation, and why keeping raw data local changes the system tradeoff.
easy
medium
medium
medium
medium