In transfer learning, which scenario best describes its use?

• A. When target data is scarce but related data exists.
• B. When there is abundant labeled data in the target domain.
• C. When you want to discard pre-trained models.
• D. When you want to build a model from scratch.

Answer: (A)