--- title: Objects that Simplify the Creation of Neural Networks framework: metalperformanceshaders role: collectionGroup role_heading: API Collection path: metalperformanceshaders/objects-that-simplify-the-creation-of-neural-networks --- # Objects that Simplify the Creation of Neural Networks Simplify the creation of neural networks using networks of filter, image, and state nodes. ## Overview Overview Graphs in Metal Performance Shaders offer a higher level graph API, intended to simplify the creation of neural networks. The graph is a network of MPSNNFilterNode, MPSNNImageNode and MPSNNStateNode objects. MPSNNImageNode represents MPSImage or MPSTemporaryImage objects MPSNNFilterNode represents MPSCNNKernel objects—each of the lower level MPSCNNKernel subclasses has an associated object that is a subclass of the MPSNNFilterNode MPSNNStateNode represents MPSState objects ## Topics ### Neural Network Graphs - [MPSNNGraph](metalperformanceshaders/mpsnngraph.md) - [MPSNNImageNode](metalperformanceshaders/mpsnnimagenode.md) - [MPSHandle](metalperformanceshaders/mpshandle.md) ### Arithmetic Layer Nodes - [MPSNNAdditionNode](metalperformanceshaders/mpsnnadditionnode.md) - [MPSNNAdditionGradientNode](metalperformanceshaders/mpsnnadditiongradientnode.md) - [MPSNNSubtractionNode](metalperformanceshaders/mpsnnsubtractionnode.md) - [MPSNNSubtractionGradientNode](metalperformanceshaders/mpsnnsubtractiongradientnode.md) - [MPSNNMultiplicationNode](metalperformanceshaders/mpsnnmultiplicationnode.md) - [MPSNNMultiplicationGradientNode](metalperformanceshaders/mpsnnmultiplicationgradientnode.md) - [MPSNNDivisionNode](metalperformanceshaders/mpsnndivisionnode.md) - [MPSNNBinaryArithmeticNode](metalperformanceshaders/mpsnnbinaryarithmeticnode.md) - [MPSNNArithmeticGradientNode](metalperformanceshaders/mpsnnarithmeticgradientnode.md) - [MPSNNArithmeticGradientStateNode](metalperformanceshaders/mpsnnarithmeticgradientstatenode.md) ### Convolution Layer Nodes - [MPSCNNBinaryConvolutionNode](metalperformanceshaders/mpscnnbinaryconvolutionnode.md) - [MPSCNNConvolutionNode](metalperformanceshaders/mpscnnconvolutionnode.md) - [MPSCNNConvolutionTransposeNode](metalperformanceshaders/mpscnnconvolutiontransposenode.md) - [MPSCNNConvolutionGradientNode](metalperformanceshaders/mpscnnconvolutiongradientnode.md) - [MPSCNNConvolutionGradientStateNode](metalperformanceshaders/mpscnnconvolutiongradientstatenode.md) - [MPSCNNCrossChannelNormalizationGradientNode](metalperformanceshaders/mpscnncrosschannelnormalizationgradientnode.md) ### Pooling Layer Nodes - [MPSCNNPoolingAverageNode](metalperformanceshaders/mpscnnpoolingaveragenode.md) - [MPSCNNDilatedPoolingMaxNode](metalperformanceshaders/mpscnndilatedpoolingmaxnode.md) - [MPSCNNPoolingL2NormNode](metalperformanceshaders/mpscnnpoolingl2normnode.md) - [MPSCNNPoolingMaxNode](metalperformanceshaders/mpscnnpoolingmaxnode.md) - [MPSCNNPoolingNode](metalperformanceshaders/mpscnnpoolingnode.md) - [MPSCNNDilatedPoolingMaxGradientNode](metalperformanceshaders/mpscnndilatedpoolingmaxgradientnode.md) - [MPSCNNPoolingAverageGradientNode](metalperformanceshaders/mpscnnpoolingaveragegradientnode.md) - [MPSCNNPoolingGradientNode](metalperformanceshaders/mpscnnpoolinggradientnode.md) - [MPSCNNPoolingL2NormGradientNode](metalperformanceshaders/mpscnnpoolingl2normgradientnode.md) - [MPSCNNPoolingMaxGradientNode](metalperformanceshaders/mpscnnpoolingmaxgradientnode.md) ### Fully Connected Layer Nodes - [MPSCNNBinaryFullyConnectedNode](metalperformanceshaders/mpscnnbinaryfullyconnectednode.md) - [MPSCNNFullyConnectedNode](metalperformanceshaders/mpscnnfullyconnectednode.md) ### Neuron Layer Nodes - [MPSCNNNeuronAbsoluteNode](metalperformanceshaders/mpscnnneuronabsolutenode.md) - [MPSCNNNeuronELUNode](metalperformanceshaders/mpscnnneuronelunode.md) - [MPSCNNNeuronHardSigmoidNode](metalperformanceshaders/mpscnnneuronhardsigmoidnode.md) - [MPSCNNNeuronLinearNode](metalperformanceshaders/mpscnnneuronlinearnode.md) - [MPSCNNNeuronPReLUNode](metalperformanceshaders/mpscnnneuronprelunode.md) - [MPSCNNNeuronReLUNNode](metalperformanceshaders/mpscnnneuronrelunnode.md) - [MPSCNNNeuronReLUNode](metalperformanceshaders/mpscnnneuronrelunode.md) - [MPSCNNNeuronSigmoidNode](metalperformanceshaders/mpscnnneuronsigmoidnode.md) - [MPSCNNNeuronSoftPlusNode](metalperformanceshaders/mpscnnneuronsoftplusnode.md) - [MPSCNNNeuronSoftSignNode](metalperformanceshaders/mpscnnneuronsoftsignnode.md) - [MPSCNNNeuronTanHNode](metalperformanceshaders/mpscnnneurontanhnode.md) - [MPSCNNNeuronExponentialNode](metalperformanceshaders/mpscnnneuronexponentialnode.md) - [MPSCNNNeuronGradientNode](metalperformanceshaders/mpscnnneurongradientnode.md) - [MPSCNNNeuronLogarithmNode](metalperformanceshaders/mpscnnneuronlogarithmnode.md) - [MPSCNNNeuronPowerNode](metalperformanceshaders/mpscnnneuronpowernode.md) - [MPSCNNNeuronNode](metalperformanceshaders/mpscnnneuronnode.md) ### Softmax Layer Nodes - [MPSCNNSoftMaxNode](metalperformanceshaders/mpscnnsoftmaxnode.md) - [MPSCNNLogSoftMaxNode](metalperformanceshaders/mpscnnlogsoftmaxnode.md) - [MPSCNNLogSoftMaxGradientNode](metalperformanceshaders/mpscnnlogsoftmaxgradientnode.md) - [MPSCNNSoftMaxGradientNode](metalperformanceshaders/mpscnnsoftmaxgradientnode.md) ### Normalization Layer Nodes - [MPSCNNCrossChannelNormalizationNode](metalperformanceshaders/mpscnncrosschannelnormalizationnode.md) - [MPSCNNLocalContrastNormalizationNode](metalperformanceshaders/mpscnnlocalcontrastnormalizationnode.md) - [MPSCNNSpatialNormalizationNode](metalperformanceshaders/mpscnnspatialnormalizationnode.md) - [MPSCNNBatchNormalizationGradientNode](metalperformanceshaders/mpscnnbatchnormalizationgradientnode.md) - [MPSCNNBatchNormalizationNode](metalperformanceshaders/mpscnnbatchnormalizationnode.md) - [MPSCNNBatchNormalizationDataSource](metalperformanceshaders/mpscnnbatchnormalizationdatasource.md) - [MPSCNNInstanceNormalizationGradientNode](metalperformanceshaders/mpscnninstancenormalizationgradientnode.md) - [MPSCNNInstanceNormalizationDataSource](metalperformanceshaders/mpscnninstancenormalizationdatasource.md) - [MPSCNNInstanceNormalizationNode](metalperformanceshaders/mpscnninstancenormalizationnode.md) - [MPSCNNLocalContrastNormalizationGradientNode](metalperformanceshaders/mpscnnlocalcontrastnormalizationgradientnode.md) - [MPSCNNSpatialNormalizationGradientNode](metalperformanceshaders/mpscnnspatialnormalizationgradientnode.md) - [MPSCNNNormalizationNode](metalperformanceshaders/mpscnnnormalizationnode.md) ### Upsampling Layer Nodes - [MPSCNNUpsamplingBilinearNode](metalperformanceshaders/mpscnnupsamplingbilinearnode.md) - [MPSCNNUpsamplingNearestNode](metalperformanceshaders/mpscnnupsamplingnearestnode.md) - [MPSCNNUpsamplingBilinearGradientNode](metalperformanceshaders/mpscnnupsamplingbilineargradientnode.md) - [MPSCNNUpsamplingNearestGradientNode](metalperformanceshaders/mpscnnupsamplingnearestgradientnode.md) ### Resampling Nodes - [MPSNNBilinearScaleNode](metalperformanceshaders/mpsnnbilinearscalenode.md) - [MPSNNLanczosScaleNode](metalperformanceshaders/mpsnnlanczosscalenode.md) - [MPSNNScaleNode](metalperformanceshaders/mpsnnscalenode.md) - [MPSImageTransformProvider](metalperformanceshaders/mpsimagetransformprovider.md) ### Dropout Layer Nodes - [MPSCNNDropoutNode](metalperformanceshaders/mpscnndropoutnode.md) - [MPSCNNDropoutGradientNode](metalperformanceshaders/mpscnndropoutgradientnode.md) ### Kernel Concatenation Nodes - [MPSNNConcatenationNode](metalperformanceshaders/mpsnnconcatenationnode.md) - [MPSNNConcatenationGradientNode](metalperformanceshaders/mpsnnconcatenationgradientnode.md) ### Loss Layer Nodes - [MPSCNNLossNode](metalperformanceshaders/mpscnnlossnode.md) - [MPSCNNYOLOLossNode](metalperformanceshaders/mpscnnyololossnode.md) - [MPSNNLabelsNode](metalperformanceshaders/mpsnnlabelsnode.md) ### Filter Node Base Classes - [MPSNNFilterNode](metalperformanceshaders/mpsnnfilternode.md) - [MPSNNGradientFilterNode](metalperformanceshaders/mpsnngradientfilternode.md) ### Protocols - [MPSNNTrainableNode](metalperformanceshaders/mpsnntrainablenode.md) ## See Also ### Neural Networks - [Training a Neural Network with Metal Performance Shaders](metalperformanceshaders/training-a-neural-network-with-metal-performance-shaders.md) - [MPSImage](metalperformanceshaders/mpsimage.md) - [MPSTemporaryImage](metalperformanceshaders/mpstemporaryimage.md) - [Convolutional Neural Network Kernels](metalperformanceshaders/convolutional-neural-network-kernels.md) - [Recurrent Neural Networks](metalperformanceshaders/recurrent-neural-networks.md)