diff options
author | DB Tsai <dbtsai@alpinenow.com> | 2014-11-25 01:57:34 -0800 |
---|---|---|
committer | Xiangrui Meng <meng@databricks.com> | 2014-11-25 01:57:34 -0800 |
commit | 89f912264603741c7d980135c26102d63e11791f (patch) | |
tree | ab0be9700b5960223e9bb1877f4d4b22e12aee05 | |
parent | 0fe54cff19759dad2dc2a0950bd6c1d31c95e858 (diff) | |
download | spark-89f912264603741c7d980135c26102d63e11791f.tar.gz spark-89f912264603741c7d980135c26102d63e11791f.tar.bz2 spark-89f912264603741c7d980135c26102d63e11791f.zip |
[SPARK-4596][MLLib] Refactorize Normalizer to make code cleaner
In this refactoring, the performance will be slightly increased due to removing
the overhead from breeze vector. The bottleneck is still in breeze norm
which is implemented by activeIterator.
This inefficiency of breeze norm will be addressed in next PR. At least,
this PR makes the code more consistent in the codebase.
Author: DB Tsai <dbtsai@alpinenow.com>
Closes #3446 from dbtsai/normalizer and squashes the following commits:
e20a2b9 [DB Tsai] first commit
-rw-r--r-- | mllib/src/main/scala/org/apache/spark/mllib/feature/Normalizer.scala | 29 |
1 files changed, 19 insertions, 10 deletions
diff --git a/mllib/src/main/scala/org/apache/spark/mllib/feature/Normalizer.scala b/mllib/src/main/scala/org/apache/spark/mllib/feature/Normalizer.scala index dfad25d57c..a9c2e23717 100644 --- a/mllib/src/main/scala/org/apache/spark/mllib/feature/Normalizer.scala +++ b/mllib/src/main/scala/org/apache/spark/mllib/feature/Normalizer.scala @@ -17,10 +17,10 @@ package org.apache.spark.mllib.feature -import breeze.linalg.{DenseVector => BDV, SparseVector => BSV, norm => brzNorm} +import breeze.linalg.{norm => brzNorm} import org.apache.spark.annotation.Experimental -import org.apache.spark.mllib.linalg.{Vector, Vectors} +import org.apache.spark.mllib.linalg.{DenseVector, SparseVector, Vector, Vectors} /** * :: Experimental :: @@ -47,22 +47,31 @@ class Normalizer(p: Double) extends VectorTransformer { * @return normalized vector. If the norm of the input is zero, it will return the input vector. */ override def transform(vector: Vector): Vector = { - var norm = brzNorm(vector.toBreeze, p) + val norm = brzNorm(vector.toBreeze, p) if (norm != 0.0) { // For dense vector, we've to allocate new memory for new output vector. // However, for sparse vector, the `index` array will not be changed, // so we can re-use it to save memory. - vector.toBreeze match { - case dv: BDV[Double] => Vectors.fromBreeze(dv :/ norm) - case sv: BSV[Double] => - val output = new BSV[Double](sv.index, sv.data.clone(), sv.length) + vector match { + case dv: DenseVector => + val values = dv.values.clone() + val size = values.size var i = 0 - while (i < output.data.length) { - output.data(i) /= norm + while (i < size) { + values(i) /= norm i += 1 } - Vectors.fromBreeze(output) + Vectors.dense(values) + case sv: SparseVector => + val values = sv.values.clone() + val nnz = values.size + var i = 0 + while (i < nnz) { + values(i) /= norm + i += 1 + } + Vectors.sparse(sv.size, sv.indices, values) case v => throw new IllegalArgumentException("Do not support vector type " + v.getClass) } } else { |