[
https://issues.apache.org/jira/browse/IGNITE-6642?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alexey Zinoviev updated IGNITE-6642:
------------------------------------
Description:
We need to be able to export/import Ignite model versions across clusters with
different versions and have exchangable & human-readable format for inference
with different systems like scikit-learn, Spark ML and etc
The PMML format is a good choice here:
PMML - Predictive Model Markup Language is XML based language which used in
SPARK MLlib and others platforms.
Here some additional info about PMML:
(i) [http://dmg.org/pmml/v4-3/GeneralStructure.html]
(i) [https://github.com/jpmml/jpmml-model]
But PMML has limitation support for Ensembles like Random Forest, Gradient
Boosted Trees, Stacking, Bagging and so on.
These cases could be covered with our own JSON format which could be easily
parsed in another system.
was:
PMML - Predictive Model Markup Language is XML based language which used in
SPARK MLlib and others platforms.
Here some additional info about PMML:
(i) [http://dmg.org/pmml/v4-3/GeneralStructure.html]
(i) [https://github.com/jpmml/jpmml-model]
> [Umbrella] Model export/import to PMML and custom JSON format
> -------------------------------------------------------------
>
> Key: IGNITE-6642
> URL: https://issues.apache.org/jira/browse/IGNITE-6642
> Project: Ignite
> Issue Type: New Feature
> Components: ml
> Reporter: Alexey Zinoviev
> Assignee: Alexey Zinoviev
> Priority: Major
> Fix For: 2.9
>
>
>
> We need to be able to export/import Ignite model versions across clusters
> with different versions and have exchangable & human-readable format for
> inference with different systems like scikit-learn, Spark ML and etc
> The PMML format is a good choice here:
> PMML - Predictive Model Markup Language is XML based language which used in
> SPARK MLlib and others platforms.
> Here some additional info about PMML:
> (i) [http://dmg.org/pmml/v4-3/GeneralStructure.html]
> (i) [https://github.com/jpmml/jpmml-model]
>
> But PMML has limitation support for Ensembles like Random Forest, Gradient
> Boosted Trees, Stacking, Bagging and so on.
> These cases could be covered with our own JSON format which could be easily
> parsed in another system.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
