A team is building a system that names the bird singing in a short recording. It follows the classical pipeline:
The model never hears the recording itself, only the three numbers. Once trained it is fixed: the same three numbers in always give the same species name out.
The model is trained on these ten labelled recordings, then asked to name each of the same ten:
| recording | highest pitch (kHz) | length (s) | notes | species |
|---|---|---|---|---|
| 1 | 4 | 2 | 5 | wren |
| 2 | 6 | 1 | 3 | warbler |
| 3 | 3 | 4 | 8 | thrush |
| 4 | 4 | 2 | 5 | wren |
| 5 | 5 | 3 | 4 | wren |
| 6 | 4 | 2 | 5 | warbler |
| 7 | 3 | 4 | 8 | wren |
| 8 | 6 | 1 | 3 | warbler |
| 9 | 4 | 2 | 5 | wren |
| 10 | 3 | 4 | 8 | thrush |
The team can pick any kind of model, make it as large as they like and train it for as long as they like.
At most, how many of the ten recordings can the trained model name correctly?
Select all that apply.