NeuralNetDenseBatch.java

/* file: NeuralNetDenseBatch.java */
/*******************************************************************************
* Copyright 2014-2019 Intel Corporation.
*
* This software and the related documents are Intel copyrighted  materials,  and
* your use of  them is  governed by the  express license  under which  they were
* provided to you (License).  Unless the License provides otherwise, you may not
* use, modify, copy, publish, distribute,  disclose or transmit this software or
* the related documents without Intel's prior written permission.
*
* This software and the related documents  are provided as  is,  with no express
* or implied  warranties,  other  than those  that are  expressly stated  in the
* License.
*******************************************************************************/

/*
 //  Content:
 //     Java example of neural network in the batch processing mode
 */

package com.intel.daal.examples.neural_networks;

import com.intel.daal.algorithms.neural_networks.*;
import com.intel.daal.algorithms.neural_networks.prediction.*;
import com.intel.daal.algorithms.neural_networks.training.*;
import com.intel.daal.algorithms.neural_networks.layers.LayerDescriptor;
import com.intel.daal.algorithms.neural_networks.layers.NextLayers;
import com.intel.daal.algorithms.neural_networks.layers.ForwardLayer;
import com.intel.daal.algorithms.neural_networks.layers.BackwardLayer;
import com.intel.daal.algorithms.neural_networks.layers.ForwardInputId;
import com.intel.daal.algorithms.neural_networks.layers.ForwardResultId;
import com.intel.daal.algorithms.neural_networks.layers.BackwardInputId;
import com.intel.daal.algorithms.neural_networks.layers.BackwardResultId;
import com.intel.daal.data_management.data.Tensor;
import com.intel.daal.data_management.data.HomogenTensor;
import com.intel.daal.data_management.data.HomogenNumericTable;
import com.intel.daal.data_management.data.NumericTable;
import com.intel.daal.data_management.data_source.DataSource;
import com.intel.daal.examples.utils.Service;
import com.intel.daal.services.DaalContext;

class NeuralNetDenseBatch {

    /* Input data set parameters */
    private static final String trainDatasetFile     = "../data/batch/neural_network_train.csv";
    private static final String trainGroundTruthFile = "../data/batch/neural_network_train_ground_truth.csv";
    private static final String testDatasetFile      = "../data/batch/neural_network_test.csv";
    private static final String testGroundTruthFile  = "../data/batch/neural_network_test_ground_truth.csv";
    private static final long batchSize              = 10;

    private static PredictionModel predictionModel;
    private static PredictionResult predictionResult;

    private static DaalContext context = new DaalContext();

    public static void main(String[] args) throws java.io.FileNotFoundException, java.io.IOException {

        trainModel();

        testModel();

        printResults();

        context.dispose();
    }

    private static void trainModel() throws java.io.FileNotFoundException, java.io.IOException {
        /* Read training data set from a .csv file and create a tensor to store input data */
        Tensor trainingData = Service.readTensorFromCSV(context, trainDatasetFile);
        Tensor trainingGroundTruth = Service.readTensorFromCSV(context, trainGroundTruthFile, true);

        /* Set learning rate for the optimization solver used in the neural network */
        double[] learningRateArray = new double[1];
        learningRateArray[0] = 0.001;
        com.intel.daal.algorithms.optimization_solver.sgd.Batch sgdAlgorithm =
            new com.intel.daal.algorithms.optimization_solver.sgd.Batch(context, Float.class, com.intel.daal.algorithms.optimization_solver.sgd.Method.defaultDense);
        sgdAlgorithm.parameter.setLearningRateSequence(new HomogenNumericTable(context, learningRateArray, 1, 1));
        sgdAlgorithm.parameter.setBatchSize(batchSize);
        sgdAlgorithm.parameter.setNIterations(trainingData.getDimensions()[0] / batchSize);

        /* Create an algorithm to compute neural network results using default method */
        TrainingBatch net = new TrainingBatch(context, sgdAlgorithm);

        long[] sampleSize = trainingData.getDimensions();
        sampleSize[0] = batchSize;

        /* Configure the neural network */
        TrainingTopology topology = NeuralNetConfigurator.configureNet(context);
        net.initialize(sampleSize, topology);

        /* Set input objects for the neural network */
        net.input.set(TrainingInputId.data, trainingData);
        net.input.set(TrainingInputId.groundTruth, trainingGroundTruth);

        /* Run the neural network training */
        TrainingResult result = net.compute();

        /* Get training and prediction models of the neural network */
        TrainingModel trainingModel = result.get(TrainingResultId.model);
        predictionModel = trainingModel.getPredictionModel(Float.class);
    }

    private static void testModel() throws java.io.FileNotFoundException, java.io.IOException {

        /* Read testing data set from a .csv file and create a tensor to store input data */
        Tensor predictionData = Service.readTensorFromCSV(context, testDatasetFile);

        /* Create an algorithm to compute the neural network predictions */
        PredictionBatch net = new PredictionBatch(context);

        long[] predictionDimensions = predictionData.getDimensions();
        net.parameter.setBatchSize(predictionDimensions[0]);

        /* Set input objects for the prediction neural network */
        net.input.set(PredictionTensorInputId.data, predictionData);
        net.input.set(PredictionModelInputId.model, predictionModel);

        /* Run the neural network prediction */
        predictionResult = net.compute();
    }

    private static void printResults() throws java.io.FileNotFoundException, java.io.IOException {

        /* Read testing ground truth from a .csv file and create a tensor to store the data */
        Tensor predictionGroundTruth = Service.readTensorFromCSV(context, testGroundTruthFile);

        /* Print results of the neural network prediction */
        Service.printTensors("Ground truth", "Neural network predictions: each class probability",
                             "Neural network classification results (first 20 observations):",
                             predictionGroundTruth, predictionResult.get(PredictionResultId.prediction), 20);
    }
}
For more complete information about compiler optimizations, see our Optimization Notice.
Select sticky button color: 
Orange (only for download buttons)