What would you like to submit? (put an 'x' inside the bracket that applies)

• question
• bug report
• feature request

Issue description

Hi @cesarsouza,

I'm using a ZeroOneLoss to evaluate a classifier with a 10-fold cross validation. During the cross validation, my holdout data often contains only 2 of n possible classes. The ZeroOneLoss assumes a binary label in such a case. When computing the loss against the predicted values, an exception is thrown if the predicted values are not binary.

For example: new ZeroOneLoss(new int[] {0, 1}).Loss(new int[] {0, 2}).

AFAICS, we need to inspect both, the expected and the actual values to determine if binary labels are used or not.