Intel DAAL AI加速——神经网络

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

# file: neural_net_dense_batch.py
#===============================================================================
# Copyright 2014-2018 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:
# !    Python example of neural network training and scoring
# !*****************************************************************************
 
#
## <a name="DAAL-EXAMPLE-PY-NEURAL_NET_DENSE_BATCH"></a>
## \example neural_net_dense_batch.py
#
 
import os
import sys
 
import numpy as np
 
from daal.algorithms.neural_networks import initializers
from daal.algorithms.neural_networks import layers
from daal.algorithms import optimization_solver
from daal.algorithms.neural_networks import training, prediction
from daal.data_management import NumericTable, HomogenNumericTable
 
utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(__file__))))
if utils_folder not in sys.path:
    sys.path.insert(0, utils_folder)
from utils import printTensors, readTensorFromCSV
 
# Input data set parameters
trainDatasetFile = os.path.join("..", "data", "batch", "neural_network_train.csv")
trainGroundTruthFile = os.path.join("..", "data", "batch", "neural_network_train_ground_truth.csv")
testDatasetFile = os.path.join("..", "data", "batch", "neural_network_test.csv")
testGroundTruthFile = os.path.join("..", "data", "batch", "neural_network_test_ground_truth.csv")
 
fc1 = 0
fc2 = 1
sm1 = 2
 
batchSize = 10
 
def configureNet():
    # Create layers of the neural network
    # Create fully-connected layer and initialize layer parameters
    fullyConnectedLayer1 = layers.fullyconnected.Batch(5)
    fullyConnectedLayer1.parameter.weightsInitializer = initializers.uniform.Batch(-0.001, 0.001)
    fullyConnectedLayer1.parameter.biasesInitializer = initializers.uniform.Batch(0, 0.5)
 
    # Create fully-connected layer and initialize layer parameters
    fullyConnectedLayer2 = layers.fullyconnected.Batch(2)
    fullyConnectedLayer2.parameter.weightsInitializer = initializers.uniform.Batch(0.5, 1)
    fullyConnectedLayer2.parameter.biasesInitializer = initializers.uniform.Batch(0.5, 1)
 
    # Create softmax layer and initialize layer parameters
    softmaxCrossEntropyLayer = layers.loss.softmax_cross.Batch()
 
    # Create configuration of the neural network with layers
    topology = training.Topology()
 
    # Add layers to the topology of the neural network
    topology.push_back(fullyConnectedLayer1)
    topology.push_back(fullyConnectedLayer2)
    topology.push_back(softmaxCrossEntropyLayer)
    topology.get(fc1).addNext(fc2)
    topology.get(fc2).addNext(sm1)
    return topology
 
 
def trainModel():
    # Read training data set from a .csv file and create a tensor to store input data
    trainingData = readTensorFromCSV(trainDatasetFile)
    trainingGroundTruth = readTensorFromCSV(trainGroundTruthFile, True)
 
    sgdAlgorithm = optimization_solver.sgd.Batch(fptype=np.float32)
 
    # Set learning rate for the optimization solver used in the neural network
    learningRate = 0.001
    sgdAlgorithm.parameter.learningRateSequence = HomogenNumericTable(1, 1, NumericTable.doAllocate, learningRate)
    # Set the batch size for the neural network training
    sgdAlgorithm.parameter.batchSize = batchSize
    sgdAlgorithm.parameter.nIterations = int(trainingData.getDimensionSize(0) / sgdAlgorithm.parameter.batchSize)
 
    # Create an algorithm to train neural network
    net = training.Batch(sgdAlgorithm)
 
    sampleSize = trainingData.getDimensions()
    sampleSize[0] = batchSize
 
    # Configure the neural network
    topology = configureNet()
    net.initialize(sampleSize, topology)
 
    # Pass a training data set and dependent values to the algorithm
    net.input.setInput(training.data, trainingData)
    net.input.setInput(training.groundTruth, trainingGroundTruth)
 
    # Run the neural network training and retrieve training model
    trainingModel = net.compute().get(training.model)
    # return prediction model
    return trainingModel.getPredictionModel_Float32()
 
 
def testModel(predictionModel):
    # Read testing data set from a .csv file and create a tensor to store input data
    predictionData = readTensorFromCSV(testDatasetFile)
 
    # Create an algorithm to compute the neural network predictions
    net = prediction.Batch()
 
    net.parameter.batchSize = predictionData.getDimensionSize(0)
 
    # Set input objects for the prediction neural network
    net.input.setModelInput(prediction.model, predictionModel)
    net.input.setTensorInput(prediction.data, predictionData)
 
    # Run the neural network prediction
    # and return results of the neural network prediction
    return net.compute()
 
 
def printResults(predictionResult):
    # Read testing ground truth from a .csv file and create a tensor to store the data
    predictionGroundTruth = readTensorFromCSV(testGroundTruthFile)
 
    printTensors(predictionGroundTruth, predictionResult.getResult(prediction.prediction),
                 "Ground truth", "Neural network predictions: each class probability",
                 "Neural network classification results (first 20 observations):", 20)
 
 
topology = ""
if __name__ == "__main__":
 
    predictionModel = trainModel()
 
    predictionResult = testModel(predictionModel)
 
    printResults(predictionResult)