Created
December 30, 2015 09:53
-
-
Save sato-cloudian/d905a9e0ce6f616bfc89 to your computer and use it in GitHub Desktop.
MyCNNMnistExample experimentation 5
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| package org.deeplearning4j.examples.convolution; | |
| import org.deeplearning4j.datasets.iterator.impl.MnistDataSetIterator; | |
| import org.deeplearning4j.eval.Evaluation; | |
| import org.deeplearning4j.nn.api.OptimizationAlgorithm; | |
| import org.deeplearning4j.nn.conf.GradientNormalization; | |
| import org.deeplearning4j.nn.conf.MultiLayerConfiguration; | |
| import org.deeplearning4j.nn.conf.NeuralNetConfiguration; | |
| import org.deeplearning4j.nn.conf.Updater; | |
| import org.deeplearning4j.nn.conf.layers.ConvolutionLayer; | |
| import org.deeplearning4j.nn.conf.layers.DenseLayer; | |
| import org.deeplearning4j.nn.conf.layers.OutputLayer; | |
| import org.deeplearning4j.nn.conf.layers.SubsamplingLayer; | |
| import org.deeplearning4j.nn.conf.layers.setup.ConvolutionLayerSetup; | |
| import org.deeplearning4j.nn.multilayer.MultiLayerNetwork; | |
| import org.deeplearning4j.nn.weights.WeightInit; | |
| import org.deeplearning4j.optimize.api.IterationListener; | |
| import org.deeplearning4j.optimize.listeners.ScoreIterationListener; | |
| import org.deeplearning4j.ui.weights.HistogramIterationListener; | |
| import org.nd4j.linalg.api.ndarray.INDArray; | |
| import org.nd4j.linalg.dataset.DataSet; | |
| import org.nd4j.linalg.dataset.SplitTestAndTrain; | |
| import org.nd4j.linalg.dataset.api.iterator.DataSetIterator; | |
| import org.nd4j.linalg.lossfunctions.LossFunctions; | |
| import org.slf4j.Logger; | |
| import org.slf4j.LoggerFactory; | |
| import java.util.ArrayList; | |
| import java.util.Arrays; | |
| import java.util.List; | |
| import java.util.Random; | |
| /** | |
| * Created by willow on 5/11/15. | |
| */ | |
| public class MyCNNMnistExample { | |
| private static final Logger log = LoggerFactory.getLogger(MyCNNMnistExample.class); | |
| public static void main(String[] args) throws Exception { | |
| int numRows = 28; | |
| int numColumns = 28; | |
| int nChannels = 1; | |
| int outputNum = 10; | |
| int numSamples = 60000; | |
| int batchSize = 100; | |
| int iterations = 1; | |
| int splitTrainNum = (int) (batchSize*.8); | |
| int seed = 123; | |
| int listenerFreq = Math.max(iterations/10, 1); | |
| DataSet mnist; | |
| SplitTestAndTrain trainTest; | |
| DataSet trainInput; | |
| List<INDArray> testInput = new ArrayList<>(); | |
| List<INDArray> testLabels = new ArrayList<>(); | |
| log.info("Load data...."); | |
| DataSetIterator mnistIter = new MnistDataSetIterator(batchSize,numSamples, true); | |
| log.info("Build model...."); | |
| MultiLayerConfiguration.Builder builder = new NeuralNetConfiguration.Builder() | |
| .seed(seed) | |
| .iterations(iterations) | |
| //.gradientNormalization(GradientNormalization.RenormalizeL2PerLayer) | |
| .optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT) | |
| .learningRate(0.001) // default | |
| //.momentum(0.9) | |
| .regularization(true) | |
| .list(3) | |
| .layer(0, new ConvolutionLayer.Builder(3, 3) // 28*28*1 => 28*28*10 | |
| .nIn(nChannels) | |
| .nOut(10) | |
| .padding(1, 1) | |
| .stride(1, 1) | |
| .weightInit(WeightInit.RELU) | |
| .activation("relu") | |
| .build()) | |
| .layer(1, new SubsamplingLayer.Builder(SubsamplingLayer.PoolingType.MAX, new int[] {2,2}) // 28*28*10 => 14*14*10 | |
| .stride(2, 2) | |
| .build()) | |
| /*.layer(1, new ConvolutionLayer.Builder(4, 4) // 14*14*10 => 7*7*20 | |
| .nIn(10) | |
| .nOut(20) | |
| .padding(2, 2) | |
| .stride(2, 2) | |
| .weightInit(WeightInit.RELU) | |
| .activation("relu") | |
| .build())*/ | |
| /*.layer(3, new SubsamplingLayer.Builder(SubsamplingLayer.PoolingType.MAX, new int[] {2,2}) // 14*14*20 => 7*7*20 | |
| .stride(2, 2) | |
| .build())*/ | |
| .layer(1, new DenseLayer.Builder().activation("relu") | |
| .nOut(200).build()) | |
| .layer(2, new OutputLayer.Builder(LossFunctions.LossFunction.RMSE_XENT) | |
| .nOut(outputNum) | |
| .weightInit(WeightInit.RELU) | |
| .activation("softmax") | |
| .updater(Updater.SGD) | |
| .build()) | |
| .backprop(true).pretrain(false); | |
| new ConvolutionLayerSetup(builder,numRows,numColumns,nChannels); | |
| MultiLayerConfiguration conf = builder.build(); | |
| MultiLayerNetwork model = new MultiLayerNetwork(conf); | |
| model.init(); | |
| log.info("Train model...."); | |
| model.setListeners(Arrays.asList((IterationListener) new ScoreIterationListener(listenerFreq), new HistogramIterationListener(listenerFreq))); | |
| while(mnistIter.hasNext()) { | |
| mnist = mnistIter.next(); | |
| trainTest = mnist.splitTestAndTrain(splitTrainNum, new Random(seed)); // train set that is the result | |
| trainInput = trainTest.getTrain(); // get feature matrix and labels for training | |
| testInput.add(trainTest.getTest().getFeatureMatrix()); | |
| testLabels.add(trainTest.getTest().getLabels()); | |
| model.fit(trainInput); | |
| } | |
| log.info("Evaluate weights...."); | |
| log.info("Evaluate model...."); | |
| Evaluation eval = new Evaluation(outputNum); | |
| for(int i = 0; i < testInput.size(); i++) { | |
| INDArray output = model.output(testInput.get(i)); | |
| eval.eval(testLabels.get(i), output); | |
| } | |
| INDArray output = model.output(testInput.get(0)); | |
| eval.eval(testLabels.get(0), output); | |
| log.info(eval.stats()); | |
| log.info("****************Example finished********************"); | |
| } | |
| } |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Examples labeled as 0 classified by model as 0: 1094 times | |
| Examples labeled as 0 classified by model as 1: 1 times | |
| Examples labeled as 0 classified by model as 2: 3 times | |
| Examples labeled as 0 classified by model as 3: 3 times | |
| Examples labeled as 0 classified by model as 4: 3 times | |
| Examples labeled as 0 classified by model as 5: 4 times | |
| Examples labeled as 0 classified by model as 6: 19 times | |
| Examples labeled as 0 classified by model as 7: 1 times | |
| Examples labeled as 0 classified by model as 8: 21 times | |
| Examples labeled as 1 classified by model as 1: 1316 times | |
| Examples labeled as 1 classified by model as 2: 8 times | |
| Examples labeled as 1 classified by model as 3: 2 times | |
| Examples labeled as 1 classified by model as 5: 1 times | |
| Examples labeled as 1 classified by model as 6: 7 times | |
| Examples labeled as 1 classified by model as 7: 1 times | |
| Examples labeled as 1 classified by model as 8: 8 times | |
| Examples labeled as 1 classified by model as 9: 2 times | |
| Examples labeled as 2 classified by model as 0: 24 times | |
| Examples labeled as 2 classified by model as 1: 22 times | |
| Examples labeled as 2 classified by model as 2: 989 times | |
| Examples labeled as 2 classified by model as 3: 18 times | |
| Examples labeled as 2 classified by model as 4: 29 times | |
| Examples labeled as 2 classified by model as 5: 3 times | |
| Examples labeled as 2 classified by model as 6: 40 times | |
| Examples labeled as 2 classified by model as 7: 20 times | |
| Examples labeled as 2 classified by model as 8: 36 times | |
| Examples labeled as 2 classified by model as 9: 8 times | |
| Examples labeled as 3 classified by model as 0: 13 times | |
| Examples labeled as 3 classified by model as 1: 26 times | |
| Examples labeled as 3 classified by model as 2: 35 times | |
| Examples labeled as 3 classified by model as 3: 1068 times | |
| Examples labeled as 3 classified by model as 4: 1 times | |
| Examples labeled as 3 classified by model as 5: 37 times | |
| Examples labeled as 3 classified by model as 6: 9 times | |
| Examples labeled as 3 classified by model as 7: 22 times | |
| Examples labeled as 3 classified by model as 8: 19 times | |
| Examples labeled as 3 classified by model as 9: 18 times | |
| Examples labeled as 4 classified by model as 0: 3 times | |
| Examples labeled as 4 classified by model as 1: 10 times | |
| Examples labeled as 4 classified by model as 2: 6 times | |
| Examples labeled as 4 classified by model as 3: 1 times | |
| Examples labeled as 4 classified by model as 4: 1024 times | |
| Examples labeled as 4 classified by model as 5: 2 times | |
| Examples labeled as 4 classified by model as 6: 28 times | |
| Examples labeled as 4 classified by model as 7: 3 times | |
| Examples labeled as 4 classified by model as 8: 13 times | |
| Examples labeled as 4 classified by model as 9: 124 times | |
| Examples labeled as 5 classified by model as 0: 21 times | |
| Examples labeled as 5 classified by model as 1: 28 times | |
| Examples labeled as 5 classified by model as 2: 4 times | |
| Examples labeled as 5 classified by model as 3: 75 times | |
| Examples labeled as 5 classified by model as 4: 15 times | |
| Examples labeled as 5 classified by model as 5: 821 times | |
| Examples labeled as 5 classified by model as 6: 23 times | |
| Examples labeled as 5 classified by model as 7: 7 times | |
| Examples labeled as 5 classified by model as 8: 41 times | |
| Examples labeled as 5 classified by model as 9: 13 times | |
| Examples labeled as 6 classified by model as 0: 10 times | |
| Examples labeled as 6 classified by model as 1: 13 times | |
| Examples labeled as 6 classified by model as 2: 15 times | |
| Examples labeled as 6 classified by model as 4: 8 times | |
| Examples labeled as 6 classified by model as 5: 19 times | |
| Examples labeled as 6 classified by model as 6: 1115 times | |
| Examples labeled as 6 classified by model as 8: 8 times | |
| Examples labeled as 6 classified by model as 9: 1 times | |
| Examples labeled as 7 classified by model as 0: 10 times | |
| Examples labeled as 7 classified by model as 1: 36 times | |
| Examples labeled as 7 classified by model as 2: 28 times | |
| Examples labeled as 7 classified by model as 3: 6 times | |
| Examples labeled as 7 classified by model as 4: 26 times | |
| Examples labeled as 7 classified by model as 5: 1 times | |
| Examples labeled as 7 classified by model as 6: 3 times | |
| Examples labeled as 7 classified by model as 7: 1079 times | |
| Examples labeled as 7 classified by model as 8: 10 times | |
| Examples labeled as 7 classified by model as 9: 65 times | |
| Examples labeled as 8 classified by model as 0: 15 times | |
| Examples labeled as 8 classified by model as 1: 43 times | |
| Examples labeled as 8 classified by model as 2: 12 times | |
| Examples labeled as 8 classified by model as 3: 63 times | |
| Examples labeled as 8 classified by model as 4: 3 times | |
| Examples labeled as 8 classified by model as 5: 39 times | |
| Examples labeled as 8 classified by model as 6: 13 times | |
| Examples labeled as 8 classified by model as 7: 7 times | |
| Examples labeled as 8 classified by model as 8: 962 times | |
| Examples labeled as 8 classified by model as 9: 32 times | |
| Examples labeled as 9 classified by model as 0: 21 times | |
| Examples labeled as 9 classified by model as 1: 25 times | |
| Examples labeled as 9 classified by model as 2: 15 times | |
| Examples labeled as 9 classified by model as 3: 26 times | |
| Examples labeled as 9 classified by model as 4: 40 times | |
| Examples labeled as 9 classified by model as 5: 8 times | |
| Examples labeled as 9 classified by model as 6: 3 times | |
| Examples labeled as 9 classified by model as 7: 33 times | |
| Examples labeled as 9 classified by model as 8: 12 times | |
| Examples labeled as 9 classified by model as 9: 1002 times | |
| ==========================Scores======================================== | |
| Accuracy: 0.871 | |
| Precision: 0.872 | |
| Recall: 0.8691 | |
| F1 Score: 0.8705482879550709 | |
| =========================================================================== |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment