Deprecation Notice: With the introduction of daal4py, a package that supersedes PyDAAL, Intel is deprecating PyDAAL and will discontinue support starting with Intel® DAAL 2021 and Intel® Distribution for Python 2021. Until then Intel will continue to provide compatible pyDAAL pip and conda packages for newer releases of Intel DAAL and make it available in open source. However, Intel will not add the new features of Intel DAAL to pyDAAL. Intel recommends developers switch to and use daal4py.

Note: To find daal4py examples, refer to daal4py documentation or browse github repository.

 # file:
 # 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:
 # !    Python example of forward and backward two-dimensional convolution layer usage
 # !
 # !*****************************************************************************
 ## \example
 import os
 import sys
 from daal.algorithms.neural_networks import layers
 from daal.data_management import HomogenTensor, TensorIface
 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 printTensor
 # Input data set name
 datasetFileName = os.path.join("..", "data", "batch", "layer.csv")
 if __name__ == "__main__":
     # Create collection of dimension sizes of the input data tensor
     inDims = [2, 1, 16, 16]
     tensorData = HomogenTensor(inDims, TensorIface.doAllocate, 1.0)
     # Create an algorithm to compute forward two-dimensional convolution layer results using default method
     convolution2dLayerForward = layers.convolution2d.forward.Batch()
     convolution2dLayerForward.input.setInput(, tensorData)
     # Compute forward two-dimensional convolution layer results
     forwardResult = convolution2dLayerForward.compute()
     printTensor(forwardResult.getResult(layers.forward.value), "Two-dimensional convolution layer result (first 5 rows):", 5, 15)
                 "Two-dimensional convolution layer weights (first 5 rows):", 5, 15)
     gDims = forwardResult.getResult(layers.forward.value).getDimensions()
     # Create input gradient tensor for backward two-dimensional convolution layer
     tensorDataBack = HomogenTensor(gDims, TensorIface.doAllocate, 0.01)
     # Create an algorithm to compute backward two-dimensional convolution layer results using default method
     convolution2dLayerBackward = layers.convolution2d.backward.Batch()
     convolution2dLayerBackward.input.setInput(layers.backward.inputGradient, tensorDataBack)
     convolution2dLayerBackward.input.setInputLayerData(layers.backward.inputFromForward, forwardResult.getResultLayerData(layers.forward.resultForBackward))
     # Compute backward two-dimensional convolution layer results
     backwardResult = convolution2dLayerBackward.compute()
                 "Two-dimensional convolution layer backpropagation gradient result (first 5 rows):", 5, 15)
                 "Two-dimensional convolution layer backpropagation weightDerivative result (first 5 rows):", 5, 15)
                 "Two-dimensional convolution layer backpropagation biasDerivative result (first 5 rows):", 5, 15)
For more complete information about compiler optimizations, see our Optimization Notice.
Select sticky button color: 
Orange (only for download buttons)