This repository contains examples of using deeplearning4j with Spark ML.
A number of examples based on the Spark Notebook:
- dl4j-iris - demonstrates Iris classification using a deep-belief network (Scala)
A number of standalone example applications:
- ml.JavaIrisClassification
- ml.JavaLfwClassification
- ml.JavaMnistClassification (* broken in dl4j rc0)
- Compile project with maven.
$ mvn clean package -Dspark.version=1.4.0 -Dhadoop.version=2.2.0
These instructions are temporary until the next release of Spark Notebook. 1 . Open the example notebook. The dl4j-spark-ml package will be automatically loaded.
Before running example application, it is necessary to set up SPARK_HOME
env variable.
$ export SPARK_HOME=<Your Spark Path>
$ bin/run-example
Usage: ./bin/run-example <example-class> [example-args]
- set MASTER=XX to use a specific master
- can use abbreviated example class name relative to org.deeplearning4j
(e.g. ml.JavaIrisClassification, ml.JavaLfwClassification)
For example,
$ bin/run-example ml.JavaIrisClassification