org.apache.sysml.runtime.matrix.data

Class LibMatrixDNN

java.lang.Object

org.apache.sysml.runtime.matrix.data.LibMatrixDNN

public class LibMatrixDNN
extends Object

Field Summary

Fields

Modifier and Type	Field and Description
static boolean	DISPLAY_STATISTICS
protected static org.apache.commons.logging.Log	LOG

Constructor Summary

Constructors

Constructor and Description
LibMatrixDNN()

Method Summary

Modifier and Type	Method and Description
static void	appendStatistics(StringBuilder sb)
static void	biasAdd(MatrixBlock input, MatrixBlock bias, MatrixBlock outputBlock, int numThreads) Performs the operation corresponding to the DML script: ones = matrix(1, rows=1, cols=Hout*Wout) output = input + matrix(bias %*% ones, rows=1, cols=F*Hout*Wout) This operation is often followed by conv2d and hence we have introduced bias_add(input, bias) built-in function
static void	biasMultiply(MatrixBlock input, MatrixBlock bias, MatrixBlock outputBlock, int numThreads) Performs the operation corresponding to the DML script: ones = matrix(1, rows=1, cols=Hout*Wout) output = input * matrix(bias %*% ones, rows=1, cols=F*Hout*Wout) This operation is often followed by conv2d and hence we have introduced bias_multiply(input, bias) built-in function
static void	conv2d(MatrixBlock input, MatrixBlock filter, MatrixBlock outputBlock, ConvolutionParameters params) This method performs convolution (i.e.
static void	conv2dBackwardData(MatrixBlock filter, MatrixBlock dout, MatrixBlock outputBlock, ConvolutionParameters params) This method computes the backpropogation errors for previous layer of convolution operation
static void	conv2dBackwardFilter(MatrixBlock input, MatrixBlock dout, MatrixBlock outputBlock, ConvolutionParameters params) This method computes the backpropogation errors for filter of convolution operation
static void	maxpooling(MatrixBlock input, MatrixBlock outputBlock, ConvolutionParameters params)
static void	maxpoolingBackward(MatrixBlock input, MatrixBlock dout, MatrixBlock outputBlock, ConvolutionParameters params, boolean performReluBackward) This method computes the backpropogation errors for previous layer of maxpooling operation
static void	reluBackward(MatrixBlock input, MatrixBlock dout, MatrixBlock outputBlock, int numThreads) This method computes the backpropagation errors for previous layer of relu operation
static void	resetStatistics()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

LOG

protected static final org.apache.commons.logging.Log LOG

DISPLAY_STATISTICS

public static boolean DISPLAY_STATISTICS

Constructor Detail

LibMatrixDNN

public LibMatrixDNN()

Method Detail

appendStatistics

public static void appendStatistics(StringBuilder sb)

resetStatistics

public static void resetStatistics()

conv2d

public static void conv2d(MatrixBlock input,
                          MatrixBlock filter,
                          MatrixBlock outputBlock,
                          ConvolutionParameters params)
                   throws DMLRuntimeException

This method performs convolution (i.e. cross-correlation) operation on input

Parameters:

input - input batch

filter - filter

outputBlock - output of convolution

params - convolution parameters

Throws:

DMLRuntimeException - if DMLRuntimeException occurs

conv2dBackwardData

public static void conv2dBackwardData(MatrixBlock filter,
                                      MatrixBlock dout,
                                      MatrixBlock outputBlock,
                                      ConvolutionParameters params)
                               throws DMLRuntimeException

This method computes the backpropogation errors for previous layer of convolution operation

Parameters:

filter - filter used in conv2d

dout - errors from next layer

outputBlock - output errors

params - convolution parameters

Throws:

DMLRuntimeException - if DMLRuntimeException occurs

conv2dBackwardFilter

public static void conv2dBackwardFilter(MatrixBlock input,
                                        MatrixBlock dout,
                                        MatrixBlock outputBlock,
                                        ConvolutionParameters params)
                                 throws DMLRuntimeException

This method computes the backpropogation errors for filter of convolution operation

Parameters:

input - input image

dout - errors from next layer

outputBlock - output errors

params - convolution parameters

Throws:

DMLRuntimeException - if DMLRuntimeException occurs

maxpoolingBackward

public static void maxpoolingBackward(MatrixBlock input,
                                      MatrixBlock dout,
                                      MatrixBlock outputBlock,
                                      ConvolutionParameters params,
                                      boolean performReluBackward)
                               throws DMLRuntimeException

This method computes the backpropogation errors for previous layer of maxpooling operation

Parameters:

input - input matrix

dout - dout matrix

outputBlock - output matrix

params - convolution parameters

performReluBackward - perform ReLU backward

Throws:

DMLRuntimeException - if DMLRuntimeException occurs

reluBackward

public static void reluBackward(MatrixBlock input,
                                MatrixBlock dout,
                                MatrixBlock outputBlock,
                                int numThreads)
                         throws DMLRuntimeException

This method computes the backpropagation errors for previous layer of relu operation

Parameters:

input - input matrix

dout - errors from next layer

outputBlock - output matrix

numThreads - number of threads

Throws:

DMLRuntimeException - if DMLRuntimeException occurs

biasAdd

public static void biasAdd(MatrixBlock input,
                           MatrixBlock bias,
                           MatrixBlock outputBlock,
                           int numThreads)
                    throws DMLRuntimeException

Performs the operation corresponding to the DML script: ones = matrix(1, rows=1, cols=Hout*Wout) output = input + matrix(bias %*% ones, rows=1, cols=F*Hout*Wout) This operation is often followed by conv2d and hence we have introduced bias_add(input, bias) built-in function

Parameters:

input - input matrix

bias - bias matrix

outputBlock - output matrix

numThreads - number of threads

Throws:

DMLRuntimeException - if DMLRuntimeException occurs

biasMultiply

public static void biasMultiply(MatrixBlock input,
                                MatrixBlock bias,
                                MatrixBlock outputBlock,
                                int numThreads)
                         throws DMLRuntimeException

Performs the operation corresponding to the DML script: ones = matrix(1, rows=1, cols=Hout*Wout) output = input * matrix(bias %*% ones, rows=1, cols=F*Hout*Wout) This operation is often followed by conv2d and hence we have introduced bias_multiply(input, bias) built-in function

Parameters:

input - input matrix

bias - bias matrix

outputBlock - output matrix

numThreads - number of threads

Throws:

DMLRuntimeException - if DMLRuntimeException occurs

maxpooling

public static void maxpooling(MatrixBlock input,
                              MatrixBlock outputBlock,
                              ConvolutionParameters params)
                       throws DMLRuntimeException

Throws:

DMLRuntimeException