試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 14 / 65
I want to improve a model without sharing highly sensitive transaction data across multiple planets. Which distributed learning strategy is most appropriate?
①Collect data from all planets and perform centralized training at headquarters.
②Encrypt the raw data, transfer it, decrypt it centrally, and train.
③Use federated learning that performs local training on each planet and updates the global model via secure aggregation.
④Publish and share anonymized raw data.
⑤Perform all updates using only synthetic data.