Instruction: Explain the criteria and metrics you would use to assess a transfer learning model's performance.
Context: Aiming to understand the candidate's knowledge on performance evaluation, this question also touches on the importance of validation in machine learning projects.
Official answer available
Preview the opening of the answer, then unlock the full walkthrough.
The way I'd approach it in an interview is this: I evaluate transfer learning by comparing it against meaningful baselines: training from scratch, fixed-feature approaches, and different degrees of fine-tuning. The question is not just whether the model works, but whether...
easy
medium
hard
hard
hard