フリー問題

問題文

A regulator asks a supplier why a handful of well chosen examples cannot establish that a medical model is reliable. The supplier lists reasons for a statistical approach. Which pair of reasons matches the syllabus?

選択肢

1. A trained model meets a shifted license in service, and a signed output can still be wrong, so a wide sample and a stated clause carry the claim.
2. A trained model meets a shifted distribution in service, and a confident output can still be wrong, so a wider sample and a stated interval carry the claim.
3. A trained model meets a fixed distribution in service, and a confident output can never be wrong, so a small sample and a stated interval carry the claim.
4. A trained model meets a shifted schedule in service, and a delayed output can still be wrong, so a wide sample and a stated deadline carry the claim.