Repository: flink Updated Branches: refs/heads/master 995f8f969 -> d951f1120
[ml] [docs] Replaces remaining CoCoA names by SVM Project: http://git-wip-us.apache.org/repos/asf/flink/repo Commit: http://git-wip-us.apache.org/repos/asf/flink/commit/d951f112 Tree: http://git-wip-us.apache.org/repos/asf/flink/tree/d951f112 Diff: http://git-wip-us.apache.org/repos/asf/flink/diff/d951f112 Branch: refs/heads/master Commit: d951f1120558dc54d0932c766aaae1416f58e81f Parents: 995f8f9 Author: Till Rohrmann <trohrm...@apache.org> Authored: Thu May 28 17:07:53 2015 +0200 Committer: Till Rohrmann <trohrm...@apache.org> Committed: Thu May 28 17:07:53 2015 +0200 ---------------------------------------------------------------------- docs/libs/ml/svm.md | 8 ++++---- .../main/scala/org/apache/flink/ml/classification/SVM.scala | 4 ++-- .../org/apache/flink/ml/classification/SVMITSuite.scala | 6 +++--- 3 files changed, 9 insertions(+), 9 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/flink/blob/d951f112/docs/libs/ml/svm.md ---------------------------------------------------------------------- diff --git a/docs/libs/ml/svm.md b/docs/libs/ml/svm.md index ec5cc6a..a9c94ec 100644 --- a/docs/libs/ml/svm.md +++ b/docs/libs/ml/svm.md @@ -59,18 +59,18 @@ The implementation of this algorithm is based on the work of ## Operations -`CoCoA` is a `Predictor`. +`SVM` is a `Predictor`. As such, it supports the `fit` and `predict` operation. ### Fit -CoCoA is trained given a set of `LabeledVector`: +SVM is trained given a set of `LabeledVector`: * `fit: DataSet[LabeledVector] => Unit` ### Predict -CoCoA predicts for all subtypes of `Vector` the corresponding class label: +SVM predicts for all subtypes of `Vector` the corresponding class label: * `predict[T <: Vector]: DataSet[T] => DataSet[LabeledVector]` @@ -125,7 +125,7 @@ The SVM implementation can be controlled by the following parameters: <td><strong>Regularization</strong></td> <td> <p> - Defines the regularization constant of the CoCoA algorithm. + Defines the regularization constant of the SVM algorithm. The higher the value, the smaller will the 2-norm of the weight vector be. In case of a SVM with hinge loss this means that the SVM margin will be wider even though it might contain some false classifications. (Default value: <strong>1.0</strong>) http://git-wip-us.apache.org/repos/asf/flink/blob/d951f112/flink-staging/flink-ml/src/main/scala/org/apache/flink/ml/classification/SVM.scala ---------------------------------------------------------------------- diff --git a/flink-staging/flink-ml/src/main/scala/org/apache/flink/ml/classification/SVM.scala b/flink-staging/flink-ml/src/main/scala/org/apache/flink/ml/classification/SVM.scala index a08fef2..a186c5d 100644 --- a/flink-staging/flink-ml/src/main/scala/org/apache/flink/ml/classification/SVM.scala +++ b/flink-staging/flink-ml/src/main/scala/org/apache/flink/ml/classification/SVM.scala @@ -68,7 +68,7 @@ import breeze.linalg.{Vector => BreezeVector, DenseVector => BreezeDenseVector} * {{{ * val trainingDS: DataSet[LabeledVector] = env.readSVMFile(pathToTrainingFile) * - * val svm = CoCoA() + * val svm = SVM() * .setBlocks(10) * .setIterations(10) * .setLocalIterations(10) @@ -103,7 +103,7 @@ import breeze.linalg.{Vector => BreezeVector, DenseVector => BreezeDenseVector} * (Default value: '''10''') * * - [[org.apache.flink.ml.classification.SVM.Regularization]]: - * Defines the regularization constant of the CoCoA algorithm. The higher the value, the smaller + * Defines the regularization constant of the SVM algorithm. The higher the value, the smaller * will the 2-norm of the weight vector be. In case of a SVM with hinge loss this means that the * SVM margin will be wider even though it might contain some false classifications. * (Default value: '''1.0''') http://git-wip-us.apache.org/repos/asf/flink/blob/d951f112/flink-staging/flink-ml/src/test/scala/org/apache/flink/ml/classification/SVMITSuite.scala ---------------------------------------------------------------------- diff --git a/flink-staging/flink-ml/src/test/scala/org/apache/flink/ml/classification/SVMITSuite.scala b/flink-staging/flink-ml/src/test/scala/org/apache/flink/ml/classification/SVMITSuite.scala index cdb4ffc..55ef056 100644 --- a/flink-staging/flink-ml/src/test/scala/org/apache/flink/ml/classification/SVMITSuite.scala +++ b/flink-staging/flink-ml/src/test/scala/org/apache/flink/ml/classification/SVMITSuite.scala @@ -30,7 +30,7 @@ class SVMITSuite extends FlatSpec with Matchers with FlinkTestBase { it should "train a SVM" in { val env = ExecutionEnvironment.getExecutionEnvironment - val cocoa = SVM(). + val svm = SVM(). setBlocks(env.getParallelism). setIterations(100). setLocalIterations(100). @@ -40,9 +40,9 @@ class SVMITSuite extends FlatSpec with Matchers with FlinkTestBase { val trainingDS = env.fromCollection(Classification.trainingData) - cocoa.fit(trainingDS) + svm.fit(trainingDS) - val weightVector = cocoa.weightsOption.get.collect().apply(0) + val weightVector = svm.weightsOption.get.collect().apply(0) weightVector.valuesIterator.zip(Classification.expectedWeightVector.valueIterator).foreach { case (weight, expectedWeight) =>
