[SPARK-16880] Improve ANN training, add training data persist if needed

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 2.0.1, 2.1.0
• Component/s: ML, MLlib
• Assignee: Weichen Xu
• Reporter: Weichen Xu
• Created: 03/Aug/16 16:18
• Updated: 04/Aug/16 20:42
• Resolved: 04/Aug/16 20:41

Description

The ANN layer training does not persist input data RDD,
so that it may cause overhead cost if the RDD need to compute from lineage.

Issue Links

• links to: [Github] Pull Request #14483 (WeichenXu123)