Question: Using the source's explanation, discuss why the similarities between internet images and user-uploaded mobile-app images allow a neural network to benefit from additional internet training data.

Sample answer: Because internet images and mobile-app images share some similarities in how cats appear, patterns the network learns from internet images (such as fur texture, shapes, or poses) are not entirely specific to that source. This overlap means the network can apply knowledge gained from internet images when classifying mobile-app images, even though the two image sets are not identical.

Key points:

• Additional internet images provide more examples of what cats do/do not look like
• Internet and mobile-app images share some similarities
• Shared similarities enable knowledge transfer between the two image sources
• The network applies knowledge acquired from internet images to mobile-app images

Rubric: Full credit for explaining that shared similarities enable knowledge transfer, that additional images provide more examples of cats, and that this transfer benefits performance on mobile-app images without introducing unsupported claims.