試験

問題 38 / 65

Which method is most suitable when multiple banks want to jointly train a loan screening model without sharing personal data?

Aggregate raw data on a central server and perform centralized training.
Banks partially exchange raw data by mutual agreement.
Train only on synthetic data and ignore real-world trends.
Use federated learning and protect with techniques such as differential privacy.
Only share the trained weights without protection and do not perform any verification.

当サイトでは、ユーザー体験の向上を目的としてCookieを使用しています。サイトの利用を継続することで、Cookieの使用に同意したものとみなされます。