Github user karlhigley commented on a diff in the pull request:
https://github.com/apache/spark/pull/6613#discussion_r32949092
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/feature/RandomProjection.scala ---
@@ -0,0 +1,148 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.mllib.feature
+
+import org.apache.spark.mllib.linalg.Matrix
+import org.apache.spark.mllib.linalg.distributed.{BlockMatrix, RowMatrix,
CoordinateMatrix, MatrixEntry}
+import org.apache.spark.rdd.RDD
+import scala.collection.mutable
+import breeze.stats.distributions.{Uniform, Binomial}
+import org.apache.spark.SparkContext
+import org.apache.spark.ml.param.{ParamMap}
+import org.apache.spark.sql.Row
+
+class RandomProjection(intrinsicDimension: Int) {
+
+ /**
+ * recursive implementation of random, non repeating list
+ * @param data
+ * @param maxValue
+ * @return
+ */
+ def nonRepeatingRandomItem(data: mutable.MutableList[Int], maxValue:
Int): Int = {
+ val rand = Uniform(0.0, maxValue).draw().toInt
+ if (data.contains(rand)) nonRepeatingRandomItem(data, maxValue)
+ else rand
+ }
+
+ /**
+ * generate list of non repeating, random values
+ * @param nonZeroRange
+ * @param maxValue
+ * @return
+ */
+ def drawNonZeroIndices(nonZeroRange: Range, maxValue: Int): List[Int] = {
+ val nonZeroIndices = mutable.MutableList[Int]()
+ nonZeroRange map { j =>
+ nonZeroIndices += nonRepeatingRandomItem(nonZeroIndices, maxValue)
+ }
+ nonZeroIndices.toList
+ }
+
+ /**
+ * draw values
+ * @param nonZeroRange
+ * @return
+ */
+ def drawNonZeroValues(nonZeroRange: Range): IndexedSeq[Double] = {
+ nonZeroRange map { _ =>
+ // random value, either -1 or 1
+ new Binomial(1, 0.5).draw().toDouble * 2 - 1
+ }
+ }
+
+ /**
+ * @param value
+ * @param density
+ * @param newDimensions
+ * @return
+ */
+ def scaleNonZeroRandomValue(value: Double, density: Double,
newDimensions: Int): Double = {
+ Math.sqrt(1 / density) / Math.sqrt(newDimensions) * value
+ }
+
+ /**
+ * in RP, one row refers to one dimension of the intrinsic
+ * @param origDimensions
+ * @param density
+ * @return
+ */
+ def computeRPRows(origDimensions: Int, density: Double):
List[MatrixEntry] = {
+ // one row for each dimension of the dataset
+ val rows = 0 until origDimensions map { rowIndex =>
+
+ /**
+ * flip coin 'origDimensions' times with a probability of 'density'
and count
+ */
+ val nonZero = new Binomial(intrinsicDimension, density).sample()
+ val nonZeroRange = 0 until nonZero
+
+ val nonZeroIndices = drawNonZeroIndices(nonZeroRange,
intrinsicDimension)
+ val nonZeroValues = drawNonZeroValues(nonZeroRange)
+
+ require(nonZeroValues.length == nonZeroIndices.length,
+ "nonZero values and indices must have same length")
+
+ val merged = nonZeroIndices.zip(nonZeroValues).map { item =>
+ val colIndex = item._1
+ // scale each value
+ val value = scaleNonZeroRandomValue(value = item._2, density,
origDimensions)
+ new MatrixEntry(rowIndex, colIndex, value)
+ }
+ merged.toList
+ }
+
+ /**
+ * make sure the matrix has the correct dimension while initializing
it with the
+ * required dimensions
+ */
+ val initialSize = List(new MatrixEntry(origDimensions - 1,
intrinsicDimension - 1, 0.0))
--- End diff --
It seems like this could introduce duplicate entries in the
CoordinateMatrix created in `computeRPMatrix()` below. CoordinateMatrix allows
the dimensions to be specified as constructor params, which might be preferable.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
