Case context: A team has 200,000 internet images and 5,000 images from mobile-app users. It can build a huge neural network and train it long enough on all 205,000 images.

Question: Based only on the source, should the team add the internet images, and what conditions justify that decision?

Sample answer: Yes. The team should add the internet images because it can use a huge neural network and train it long enough on the complete combined dataset. In theory, those conditions mean there is no harm in trying to make the algorithm perform well on both sources, despite the 40:1 dataset-size ratio.

Key points:

• Add the internet images.
• Use a huge neural network.
• Train long enough on all 205,000 images.
• The datasets have a 40:1 size ratio.
• There is no theoretical harm in pursuing performance on both sources under these conditions.

Rubric: Full credit requires a clear decision to add the images, both conditions of a huge network and long training, acknowledgment of learning from both sources, and correct treatment of the 40:1 ratio as compatible with the theoretical claim.