A Spark machine learning package containing the implementation of classical Convolutional Neural Network.
This documentation is for Spark 1.3+. Other version will probably work yet not tested.
mCNN supports training of a Convolutional Neural Network. Currently ConvolutionLayer, MeanPoolingLayer are included. MaxPooling and SumPooling are under test.
A version compatible to the ANN interface in Spark 1.5 is also under development in the communityInterface folder.
// training for Mnist data set.
val topology = new CNNTopology
topology.addLayer(CNNLayer.buildConvolutionLayer(1, 6, new Scale(5, 5)))
topology.addLayer(CNNLayer.buildMeanPoolingLayer(new Scale(2, 2)))
topology.addLayer(CNNLayer.buildConvolutionLayer(6, 12, new Scale(5, 5)))
topology.addLayer(CNNLayer.buildMeanPoolingLayer(new Scale(2, 2)))
topology.addLayer(CNNLayer.buildConvolutionLayer(12, 12, new Scale(4, 4)))
val cnn: CNN = new CNN(topology).setMaxIterations(1000).setMiniBatchSize(16)
cnn.trainOneByOne(data)