MPSCNNSubtractGradient(3) MetalPerformanceShaders.framework MPSCNNSubtractGradient(3)

MPSCNNSubtractGradient

#import <MPSCNNMath.h>

Inherits MPSCNNArithmeticGradient.


(nonnull instancetype) - initWithDevice:isSecondarySourceFilter:

This depends on Metal.framework. Specifies the subtraction gradient operator. This arithmetic gradient filter requires the following inputs: gradient image from the previous layer (going backwards) and either the primary or the secondary source image from the forward pass. You will need a separate filter for the primary and secondary source images.

Without broadcasting, the arithmetic subtract gradient operation for the primary source image is a copy operation on the input gradient image (for x - y, d/dx(x - y) = 1). This copy operation can be optimized away by the graph interface.

For the secondary source image, the result is a negation of the gradient image from the previous layer (for x - y, d/dy(x - y) = -1).

Setting the broadcasting parameters results in a reduction operation (sum) across all of the applicable broadcasting dimensions (rows, columns, feature channels, or any combination thereof) to produce the destination image of the size that matches the primary/secondary input images used in the forward pass.

- (nonnull instancetype) initWithDevice: (nonnull id< MTLDevice >) device(BOOL) isSecondarySourceFilter

Initialize the subtraction gradient operator.

Parameters:

device The device the filter will run on.
isSecondarySourceFilter A boolean indicating whether the arithmetic gradient filter is operating on the primary or secondary source image from the forward pass.

Returns:

A valid MPSCNNSubtractGradient object or nil, if failure.

Reimplemented from MPSCNNArithmeticGradient.

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

Mon Jul 9 2018 Version MetalPerformanceShaders-119.3