Outlier detection identifies unusual examples within a training dataset. It is used to find observations that deviate from the expected distribution and may reflect errors, rare events, or data points requiring special handling.
The process of identifying outliers in a training set. Contrast with novelty detection.