classification

measure the accuracy/error-rate of the model:

• accuracy - the proportion of examples for which the model produces the correct class label)
• error-rate - the proportion of examples for which the model produces an incorrect class label

probability distribution estimation

use a performance metric that gives the model a continuous-valued score for each example (e.g. average log-probability the model assigns to some examples)

transcription

either:

• accuracy of transcribing entire sequences
• fine-grained measure that gives partial credit for getting some elements of sequence correct

more performance measures