[SPARK-1655] In naive Bayes, store conditional probabilities distributively.

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Auto Closed
• Component/s: MLlib
• Labels: bulk-closed
• Assignee: Aaron Staple
• Reporter: Xiangrui Meng
• Created: 28/Apr/14 02:54
• Updated: 06/Jun/19 13:57
• Resolved: 06/Jun/19 13:57

Description

In the current implementation, we collect all conditional probabilities to the driver node. When there are many labels and many features, this puts heavy load on the driver. For scalability, we should provide a way to store conditional probabilities distributively.

Issue Links

• links to: [Github] Pull Request #2491 (staple)