SystemML Examples

systemml-examples.md

MNIST LeNet

import org.apache.sysml.api.mlcontext._
val ml = new MLContext(spark)
val clf = ml.nn.examples.Mnist_lenet
val dummy = clf.generate_dummy_data
val dummyVal = clf.generate_dummy_data
val params = clf.train(dummy.X, dummy.Y, dummyVal.X, dummyVal.Y, dummy.C, dummy.Hin, dummy.Win, 1)
val probs = clf.predict(dummy.X, dummy.C, dummy.Hin, dummy.Win, params.W1, params.b1, params.W2, params.b2, params.W3, params.b3, params.W4, params.b4)
val perf = clf.eval(probs, dummy.Y)

clf.eval__docs
clf.eval__source

CSV to DataFrame on MLContext

@Test
public void testing() {
	List<StructField> fields = new ArrayList<StructField>();
	fields.add(DataTypes.createStructField("C1", DataTypes.IntegerType, true));
	fields.add(DataTypes.createStructField("C2", DataTypes.IntegerType, true));
	fields.add(DataTypes.createStructField("C3", DataTypes.IntegerType, true));
	StructType schema = DataTypes.createStructType(fields);

	Dataset<Row> df = spark.read().option("header", "true").schema(schema).csv("tongue.csv");

	Script script = dml("Y = t(X1) %*% X1;");
	MatrixMetadata mm = new MatrixMetadata(80, 3);
	script.in("X1", df, mm).out("Y");
	MLResults res = ml.execute(script);
	res.getDataFrame("Y").sort("__INDEX").show();
}