search

amplab / keystone

Simplifying robust end-to-end machine learning on Apache Spark.
http://keystone-ml.org/
Apache License 2.0
467 stars 116 forks source link

KMeans #50

Open etrain opened 9 years ago

etrain commented 9 years ago

Call into MLlib

ktham commented 9 years ago

Hi, I'm getting confused as to the relationship between MLLib and Keystone?

etrain commented 9 years ago

KeystoneML is a project designed to help people construct end-to-end machine learning pipelines at scale (e.g. do multiple levels of feature extraction before calling into a supervised learning algorithm) while MLlib is a collection of well-tested machine learning algorithms that are optimized to run fast on Spark. The projects are complimentary and we call into MLlib where appropriate.

KeystoneML is a research project to help us understand the right way to build systems that support large scale machine learning - it's not yet "production ready" but we are happy to help users who are interested in trying it out!