An online shop files every listing under one of five departments, which its database stores as codes:
| code | department |
|---|---|
| Books | |
| Toys | |
| Shoes | |
| Kitchen | |
| Garden |
A developer builds a network to file new listings automatically. To keep it simple, the network has a single output, trained with mean squared error against each listing's code. To file a listing, that output is rounded to the nearest code.
Cookbooks are the awkward case. In the training data, of cookbook listings are filed under Books and under Kitchen. Cookbooks all look alike to the network, so it ends up giving every cookbook the same output.
Training runs until the MSE is as low as the network can make it. Which statement about the result is correct?
Select all that apply.