In the context of enhancing model performance through feature selection in your machine learning project, which Databricks MLlib-supported technique is adept at identifying the most pertinent features from a vast dataset?