Hi Alex and Felix, The MLP implementation is located in the MapReduce-Legacy package:
https://github.com/apache/mahout/tree/master/mrlegacy/src/main/java/org/apache/mahout/classifier/mlp You should be able to run the MLP from the command line as follows: $ bin/mahout org.apache.mahout.classifier.mlp.TrainMultilayerPerceptron $ bin/mahout org.apache.mahout.classifier.mlp.RunMultilayerPerceptron this willl show you the usage and CLI options for each class. To get an idea of the options that can be used to used to train and test an MLP on a dataset, you could take a look at the iris dataset case in the Unit Tests: https://github.com/apache/mahout/blob/master/mrlegacy/src/test/java/org/apache/mahout/classifier/mlp/TrainMultilayerPerceptronTest.java -Specifically the args String arrays. The iris dataset is located at: $MAHOUT_HOME/mrlegacy/src/test/resources/iris.csv ex. $ mahout org.apache.mahout.classifier.mlp.TrainMultilayerPerceptron -i ./mrlegacy/src/test/resources/iris.csv -sh -labels setosa versicolor virginica -mo /tmp/model.model -ls 4 8 3 -l 0.2 -m 0.35 -r 0.0001 Andy > Date: Sun, 15 Jun 2014 12:44:59 +0200 > From: [email protected] > To: [email protected]; [email protected] > Subject: Alex and Felix are Working on MAHOUT-1551 > > Hi all! > > We are Alex and Felix from Sebastian Schelter's course at TU Berlin. In > the scope of this course we are going to work on MAHOUT-1551: Add > document to describe how to use mlp with command line > > We would be glad if someone could give us a quick update on the state of > the MLP implementation, especially because it doesn't show up on the > list of available programs (when running mahout from command line) and > we couldn't find a description of a working command line implementation > (the description in MAHOUT-1388 doesn't seem to work in the trunk). > > Thanks for your help and we are looking forward to creating a nice piece > of documentation! > > Alex & Felix
