試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 65 / 65
When improving a global AI model, which approach is most appropriate for keeping each country's data local and minimizing cross-border data transfers?
①Aggregate raw data from various locations to a central location for training.
②Generate synthetic data and train centrally.
③Transfer with only differential privacy applied.
④Encrypt everything with homomorphic encryption and perform centralized training.
⑤Federated learning (training at each site with only model updates shared)