MPSRNNSingleGateDescriptor(3) MetalPerformanceShaders.framework MPSRNNSingleGateDescriptor(3)

MPSRNNSingleGateDescriptor

#import <MPSRNNLayer.h>

Inherits MPSRNNDescriptor.


(nonnull instancetype) + createRNNSingleGateDescriptorWithInputFeatureChannels:outputFeatureChannels:


id< MPSCNNConvolutionDataSource > inputWeights
id< MPSCNNConvolutionDataSource > recurrentWeights

This depends on Metal.framework The MPSRNNSingleGateDescriptor specifies a simple recurrent block/layer descriptor. The RNN layer initialized with a MPSRNNSingleGateDescriptor transforms the input data (image or matrix), and previous output with a set of filters, each producing one feature map in the new output data. The user may provide the RNN unit a single input or a sequence of inputs.

Description of operation:

Let x_j be the input data (at time index t of sequence, j index containing quadruplet: batch index, x,y and feature index (x=y=0 for matrices)). Let h0_j be the recurrent input (previous output) data from previous time step (at time index t-1 of sequence). Let h1_i be the output data produced at this time step.

Let W_ij, U_ij be the weights for input and recurrent input data respectively Let b_i be a bias term

Let gi(x) be a neuron activation function

Then the new output image h1_i data is computed as follows:

h1_i = gi( W_ij * x_j + U_ij * h0_j  + b_i )

The '*' stands for convolution (see MPSRNNImageInferenceLayer) or matrix-vector/matrix multiplication (see MPSRNNMatrixInferenceLayer). Summation is over index j (except for the batch index), but there is no summation over repeated index i - the output index. Note that for validity all intermediate images have to be of same size and the U matrix has to be square (ie. outputFeatureChannels == inputFeatureChannels in those). Also the bias terms are scalars wrt. spatial dimensions.

Creates a MPSRNNSingleGateDescriptor

Parameters:

inputFeatureChannels The number of feature channels in the input image/matrix. Must be >= 1.
outputFeatureChannels The number of feature channels in the output image/matrix. Must be >= 1.

Returns:

A valid MPSRNNSingleGateDescriptor object or nil, if failure.

- inputWeights [read], [write], [nonatomic], [retain]

Contains weights 'W_ij', bias 'b_i' and neuron 'gi' from the simple RNN layer formula. If nil then assumed zero weights, bias and no neuron (identity mapping). Defaults to nil.

- recurrentWeights [read], [write], [nonatomic], [retain]

Contains weights 'U_ij' from the simple RNN layer formula. If nil then assumed zero weights. Defaults to nil.

Generated automatically by Doxygen for MetalPerformanceShaders.framework from the source code.

Mon Jul 9 2018 Version MetalPerformanceShaders-119.3