When training a classification model with Databricks AutoML, your dataset includes several columns that are not relevant to the classification task. Which parameter should you use to specify columns that AutoML should ignore during its calculations?