Java MultiLayerConfiguration.getInputPreProcess方法代码示例

import org.deeplearning4j.nn.conf.MultiLayerConfiguration; //导入方法依赖的package包/类
@Test
public void testDeconvolution() {

    MultiLayerConfiguration.Builder builder = new NeuralNetConfiguration.Builder().list()
            .layer(0, new Deconvolution2D.Builder(2, 2).padding(0, 0).stride(2, 2).nIn(1).nOut(3).build()) //(28-2+0)/2+1 = 14
            .layer(1, new SubsamplingLayer.Builder().kernelSize(2, 2).padding(1, 1).stride(2, 2).build()) //(14-2+2)/2+1 = 8 -> 8x8x3
            .layer(2, new OutputLayer.Builder().nOut(3).build())
            .setInputType(InputType.convolutional(28, 28, 1));

    MultiLayerConfiguration conf = builder.build();

    assertNotNull(conf.getInputPreProcess(2));
    assertTrue(conf.getInputPreProcess(2) instanceof CnnToFeedForwardPreProcessor);
    CnnToFeedForwardPreProcessor proc = (CnnToFeedForwardPreProcessor) conf.getInputPreProcess(2);
    assertEquals(8, proc.getInputHeight());
    assertEquals(8, proc.getInputWidth());
    assertEquals(3, proc.getNumChannels());

    assertEquals(8 * 8 * 3, ((FeedForwardLayer) conf.getConf(2).getLayer()).getNIn());
}
 

import org.deeplearning4j.nn.conf.MultiLayerConfiguration; //导入方法依赖的package包/类
@Test
public void testSubSamplingWithPadding() {

    MultiLayerConfiguration.Builder builder = new NeuralNetConfiguration.Builder().list()
                    .layer(0, new ConvolutionLayer.Builder(2, 2).padding(0, 0).stride(2, 2).nIn(1).nOut(3).build()) //(28-2+0)/2+1 = 14
                    .layer(1, new SubsamplingLayer.Builder().kernelSize(2, 2).padding(1, 1).stride(2, 2).build()) //(14-2+2)/2+1 = 8 -> 8x8x3
                    .layer(2, new OutputLayer.Builder().nOut(3).build())
                    .setInputType(InputType.convolutional(28, 28, 1));

    MultiLayerConfiguration conf = builder.build();

    assertNotNull(conf.getInputPreProcess(2));
    assertTrue(conf.getInputPreProcess(2) instanceof CnnToFeedForwardPreProcessor);
    CnnToFeedForwardPreProcessor proc = (CnnToFeedForwardPreProcessor) conf.getInputPreProcess(2);
    assertEquals(8, proc.getInputHeight());
    assertEquals(8, proc.getInputWidth());
    assertEquals(3, proc.getNumChannels());

    assertEquals(8 * 8 * 3, ((FeedForwardLayer) conf.getConf(2).getLayer()).getNIn());
}
 

import org.deeplearning4j.nn.conf.MultiLayerConfiguration; //导入方法依赖的package包/类
@Test
public void testUpsampling() {

    MultiLayerConfiguration.Builder builder = new NeuralNetConfiguration.Builder().list()
            .layer(new ConvolutionLayer.Builder(2, 2).padding(0, 0).stride(2, 2).nIn(1).nOut(3).build()) //(28-2+0)/2+1 = 14
            .layer(new Upsampling2D.Builder().size(3).build()) // 14 * 3 = 42!
            .layer(new OutputLayer.Builder().nOut(3).build())
            .setInputType(InputType.convolutional(28, 28, 1));

    MultiLayerConfiguration conf = builder.build();

    assertNotNull(conf.getInputPreProcess(2));
    assertTrue(conf.getInputPreProcess(2) instanceof CnnToFeedForwardPreProcessor);
    CnnToFeedForwardPreProcessor proc = (CnnToFeedForwardPreProcessor) conf.getInputPreProcess(2);
    assertEquals(42, proc.getInputHeight());
    assertEquals(42, proc.getInputWidth());
    assertEquals(3, proc.getNumChannels());

    assertEquals(42 * 42 * 3, ((FeedForwardLayer) conf.getConf(2).getLayer()).getNIn());
}
 

import org.deeplearning4j.nn.conf.MultiLayerConfiguration; //导入方法依赖的package包/类
@Test
public void testSeparableConv2D() {

    MultiLayerConfiguration.Builder builder = new NeuralNetConfiguration.Builder().list()
            .layer( new SeparableConvolution2D.Builder(2, 2)
                    .depthMultiplier(2)
                    .padding(0, 0)
                    .stride(2, 2).nIn(1).nOut(3).build()) //(28-2+0)/2+1 = 14
            .layer( new SubsamplingLayer.Builder().kernelSize(2, 2).padding(1, 1).stride(2, 2).build()) //(14-2+2)/2+1 = 8 -> 8x8x3
            .layer(2, new OutputLayer.Builder().nOut(3).build())
            .setInputType(InputType.convolutional(28, 28, 1));

    MultiLayerConfiguration conf = builder.build();

    assertNotNull(conf.getInputPreProcess(2));
    assertTrue(conf.getInputPreProcess(2) instanceof CnnToFeedForwardPreProcessor);
    CnnToFeedForwardPreProcessor proc = (CnnToFeedForwardPreProcessor) conf.getInputPreProcess(2);
    assertEquals(8, proc.getInputHeight());
    assertEquals(8, proc.getInputWidth());
    assertEquals(3, proc.getNumChannels());

    assertEquals(8 * 8 * 3, ((FeedForwardLayer) conf.getConf(2).getLayer()).getNIn());
}
 

import org.deeplearning4j.nn.conf.MultiLayerConfiguration; //导入方法依赖的package包/类
@Test
public void testDeconv2D() {

    MultiLayerConfiguration.Builder builder = new NeuralNetConfiguration.Builder().list()
            .layer( new Deconvolution2D.Builder(2, 2)
                    .padding(0, 0)
                    .stride(2, 2).nIn(1).nOut(3).build()) //(28-2+0)/2+1 = 14
            .layer( new SubsamplingLayer.Builder().kernelSize(2, 2).padding(1, 1).stride(2, 2).build()) //(14-2+2)/2+1 = 8 -> 8x8x3
            .layer(2, new OutputLayer.Builder().nOut(3).build())
            .setInputType(InputType.convolutional(28, 28, 1));

    MultiLayerConfiguration conf = builder.build();

    assertNotNull(conf.getInputPreProcess(2));
    assertTrue(conf.getInputPreProcess(2) instanceof CnnToFeedForwardPreProcessor);
    CnnToFeedForwardPreProcessor proc = (CnnToFeedForwardPreProcessor) conf.getInputPreProcess(2);
    assertEquals(8, proc.getInputHeight());
    assertEquals(8, proc.getInputWidth());
    assertEquals(3, proc.getNumChannels());

    assertEquals(8 * 8 * 3, ((FeedForwardLayer) conf.getConf(2).getLayer()).getNIn());
}
 

import org.deeplearning4j.nn.conf.MultiLayerConfiguration; //导入方法依赖的package包/类
@Test
public void testCnnToDense() {
    MultiLayerConfiguration conf =
            new NeuralNetConfiguration.Builder()
                    .list().layer(0,
                    new org.deeplearning4j.nn.conf.layers.ConvolutionLayer.Builder(
                            4, 4) // 28*28*1 => 15*15*10
                            .nIn(1).nOut(10).padding(2, 2)
                            .stride(2, 2)
                            .weightInit(WeightInit.RELU)
                            .activation(Activation.RELU)
                            .build())
                    .layer(1, new org.deeplearning4j.nn.conf.layers.DenseLayer.Builder()
                            .activation(Activation.RELU).nOut(200).build())
                    .layer(2, new OutputLayer.Builder(LossFunctions.LossFunction.MCXENT).nIn(200)
                            .nOut(5).weightInit(WeightInit.RELU)
                            .activation(Activation.SOFTMAX).build())
                    .setInputType(InputType.convolutionalFlat(28, 28, 1)).backprop(true)
                    .pretrain(false).build();

    assertNotNull(conf.getInputPreProcess(0));
    assertNotNull(conf.getInputPreProcess(1));

    assertTrue(conf.getInputPreProcess(0) instanceof FeedForwardToCnnPreProcessor);
    assertTrue(conf.getInputPreProcess(1) instanceof CnnToFeedForwardPreProcessor);

    FeedForwardToCnnPreProcessor ffcnn = (FeedForwardToCnnPreProcessor) conf.getInputPreProcess(0);
    CnnToFeedForwardPreProcessor cnnff = (CnnToFeedForwardPreProcessor) conf.getInputPreProcess(1);

    assertEquals(28, ffcnn.getInputHeight());
    assertEquals(28, ffcnn.getInputWidth());
    assertEquals(1, ffcnn.getNumChannels());

    assertEquals(15, cnnff.getInputHeight());
    assertEquals(15, cnnff.getInputWidth());
    assertEquals(10, cnnff.getNumChannels());

    assertEquals(15 * 15 * 10, ((FeedForwardLayer) conf.getConf(1).getLayer()).getNIn());
}
 

import org.deeplearning4j.nn.conf.MultiLayerConfiguration; //导入方法依赖的package包/类
/**
 * Convert a MultiLayerNetwork to a ComputationGraph
 *
 * @return ComputationGraph equivalent to this network (including parameters and updater state)
 */
public static ComputationGraph toComputationGraph(MultiLayerNetwork net) {

    //We rely heavily here on the fact that the topological sort order - and hence the layout of parameters - is
    // by definition the identical for a MLN and "single stack" computation graph. This also has to hold
    // for the updater state...

    ComputationGraphConfiguration.GraphBuilder b = new NeuralNetConfiguration.Builder()
            .graphBuilder();

    MultiLayerConfiguration origConf = net.getLayerWiseConfigurations().clone();


    int layerIdx = 0;
    String lastLayer = "in";
    b.addInputs("in");
    for (NeuralNetConfiguration c : origConf.getConfs()) {
        String currLayer = String.valueOf(layerIdx);

        InputPreProcessor preproc = origConf.getInputPreProcess(layerIdx);
        b.addLayer(currLayer, c.getLayer(), preproc, lastLayer);

        lastLayer = currLayer;
        layerIdx++;
    }
    b.setOutputs(lastLayer);

    ComputationGraphConfiguration conf = b.build();

    ComputationGraph cg = new ComputationGraph(conf);
    cg.init();

    cg.setParams(net.params());

    //Also copy across updater state:
    INDArray updaterState = net.getUpdater().getStateViewArray();
    if (updaterState != null) {
        cg.getUpdater().getUpdaterStateViewArray()
                .assign(updaterState);
    }

    return cg;
}