Select the statement(s) that represent why in certain cases, accuracy can be excluded as a performance metric for classification evaluation.Select one or more:
选项：

A: Measuring accuracy is sometimes difficult
B: The soft-margin classifier in evaluation will separate the data
C: Accuracy does not calculate the opposite of an error
D: High accuracy model sometimes would not imply it is a good model for an imbalanced dataset
E: Accuracy is sometimes not relevant for critical model