##VGG16 model for Keras
This is the Keras model of the 16-layer network used by the VGG team in the ILSVRC-2014 competition.
It has been obtained by directly converting the Caffe model provived by the authors.
Details about the network architecture can be found in the following arXiv paper:
Very Deep Convolutional Networks for Large-Scale Image Recognition
K. Simonyan, A. Zisserman
| package org.deeplearning4j.rl4j; | |
| import org.deeplearning4j.nn.api.OptimizationAlgorithm; | |
| import org.deeplearning4j.nn.conf.MultiLayerConfiguration; | |
| import org.deeplearning4j.nn.conf.NeuralNetConfiguration; | |
| import org.deeplearning4j.nn.conf.Updater; | |
| import org.deeplearning4j.nn.conf.layers.DenseLayer; | |
| import org.deeplearning4j.nn.conf.layers.OutputLayer; | |
| import org.deeplearning4j.nn.multilayer.MultiLayerNetwork; | |
| import org.deeplearning4j.nn.weights.WeightInit; |
| conf = new NeuralNetConfiguration.Builder() | |
| .seed(seed) | |
| .iterations(1) | |
| .optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT) | |
| .learningRate(0.1) | |
| .updater(Updater.NESTEROVS) | |
| .momentum(0.9) | |
| .list() | |
| .layer(0, new DenseLayer.Builder().nIn(258).nOut(1000) | |
| .weightInit(WeightInit.XAVIER) |
| Nd4j.ENFORCE_NUMERICAL_STABILITY = true; | |
| LayerFactory layerFactory = LayerFactories.getFactory(ConvolutionDownSampleLayer.class); | |
| int batchSize = 1000; | |
| /** | |
| * | |
| */ | |
| Nd4j.MAX_ELEMENTS_PER_SLICE = Integer.MAX_VALUE; | |
| Nd4j.MAX_ELEMENTS_PER_SLICE = Integer.MAX_VALUE; | |
| MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder() | |
| .optimizationAlgo(OptimizationAlgorithm.LBFGS).momentum(0.9) |
| MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder() | |
| .seed(seed) | |
| .gradientNormalization(GradientNormalization.ClipElementWiseAbsoluteValue) | |
| .gradientNormalizationThreshold(1.0) | |
| .iterations(iterations) | |
| .momentum(0.5) | |
| .momentumAfter(Collections.singletonMap(3, 0.9)) | |
| .optimizationAlgo(OptimizationAlgorithm.CONJUGATE_GRADIENT) | |
| .list(4) | |
| .layer(0, new RBM.Builder().nIn(numRows*numColumns).nOut(500) |
| MultiLayerNetwork net; | |
| //two hidden layers of 3 neurons each | |
| final int[] LSTMLayers = new int[]{3,3}; | |
| NeuralNetConfiguration.ListBuilder list = new NeuralNetConfiguration.Builder() | |
| .optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT).iterations(1) | |
| .learningRate(learningRate) | |
| .regularization(true).l2(0.0000001) | |
| .seed(76692) | |
| .weightInit(WeightInit.XAVIER) | |
| .updater(Updater.ADAM).adamMeanDecay(0.99).adamVarDecay(0.9999) |
| RandomGenerator gen = new MersenneTwister(123); | |
| MultiLayerConfiguration encoderConf = new NeuralNetConfiguration.Builder().weightInit(WeightInit.VI) | |
| .layerFactory(LayerFactories.getFactory(RBM.class)) | |
| .iterations(5) | |
| .lossFunction(LossFunctions.LossFunction.RECONSTRUCTION_CROSSENTROPY).rng(gen) | |
| .learningRate(1e-1).nIn(784).nOut(30).list(3) | |
| .hiddenLayerSizes(new int[]{2000, 1000, 500}) | |
| .build(); |
| package org.deeplearning4j.models.featuredetectors.rbm; | |
| import static org.junit.Assert.*; | |
| import org.deeplearning4j.linalg.api.ndarray.INDArray; | |
| import org.deeplearning4j.linalg.factory.NDArrays; | |
| import org.deeplearning4j.linalg.lossfunctions.LossFunctions; | |
| import org.deeplearning4j.nn.api.Model; | |
| import org.deeplearning4j.nn.conf.NeuralNetConfiguration; | |
| import org.junit.Test; |
| MultiLayerConfiguration conf = new NeuralNetConfiguration.Builder() | |
| .iterations(1) | |
| .weightInit(WeightInit.XAVIER) | |
| .activation("relu") | |
| .optimizationAlgo(OptimizationAlgorithm.STOCHASTIC_GRADIENT_DESCENT) | |
| .learningRate(0.05) | |
| // ... other hyperparameters | |
| .backprop(true) | |
| .build(); |