Answer it
Reducing dimensions in leaf photos
4 marksYou mark it yourself against the marking points
A biology club wants a model that identifies tree species from photos of leaves. Each photo is 100 by 100 pixels in greyscale, so each one is described by 10,000 numbers. The club has only 800 labelled photos.
(a) Explain why having 10,000 features for only 800 examples is a problem for training. [2]
(b) Describe, at a conceptual level, how a dimensionality reduction technique such as principal component analysis (PCA) could help. [2]
[4]