Overview of the DeepLearning Package
The DeepLearning package is a collection of tools for machine learning. The package supports several common operations used with neural networks, including classification and regression.
DeepLearning TypesCommands for Managing TensorsCommands for Managing Dataflow GraphsCommands for Constructing EstimatorsCommands for Constructing Feature ColumnsCommands for Managing SessionsDetailsCompatibility
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk0">DeepLearning Types</Text-field>
DeepLearning makes use of the following custom typesDataflowGraphEstimatorFeatureColumnLayerModelModelModelModelModelOptimizerSessionTensor
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk1">Commands for Managing Tensors</Text-field>
The core object in a DeepLearning computation is a Tensor. The following commands construct Tensor objects in the active graph.AddMultipleApplyOperationClassifyConcatenateConstantEinsteinSummationFillGetEagerExecutionGetVariableIdentityMatrixOneHotOnesOptimizerPlaceholderRandomTensorSequentialSetEagerExecutionVariableVariablesInitializerZeros
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk2">Commands for Managing Dataflow Graphs</Text-field>
The following commands allow querying and modification of the DataflowGraph in which the current computation occurs.GetDefaultGraphResetDefaultGraphRestoreSaveSetRandomSeedVariables
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk3">Commands for Constructing Estimators</Text-field>
The following commands construct Estimator objects for classification and regression tasks.DNNClassifierDNNLinearCombinedClassifierDNNLinearCombinedRegressorDNNRegressorEstimatorLinearClassifierLinearRegressor
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk4">Commands for Constructing Feature Columns</Text-field>
The following commands construct FeatureColumn objects for use with an Estimator.BucketizedColumnCategoricalColumnNumericColumn
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk5">Commands for Managing Sessions</Text-field>
The following commands manage Session objects.GetDefaultSessionSession
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk6">Details</Text-field>
The DeepLearning package is implemented using Google TensorFlow\342\204\242 and provides access to a subset of the TensorFlow Python API, version 2.2.0.
The DeepLearning package is currently not supported on the following platforms: Macs powered by Apple's M1 chip (Apple Silicon).
For Windows, a processor with AVX instructions is required. For more information, see the Release 1.6.0 section in https://github.com/tensorflow/tensorflow/blob/r1.10/RELEASE.md.
<Text-field style="Heading 2" layout="Heading 2" bookmark="compatibility">Compatibility</Text-field>
The DeepLearning package was introduced in Maple 2018.
For more information on Maple 2018 changes, see Updates in Maple 2018.See AlsoStatistics