edu.hitsz.c102c.cnn

Class Layer

java.lang.Object

edu.hitsz.c102c.cnn.Layer

All Implemented Interfaces:

java.io.Serializable

public class Layer
extends java.lang.Object
implements java.io.Serializable

CNN network layer.

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	Layer.Size Get sample size, only sampling layer has scaleSize, other layers are not null.

Method Summary

Modifier and Type	Method and Description
static Layer	buildConvLayer(int outMapNum, Layer.Size kernelSize) Build convolution layer.
static Layer	buildInputLayer(Layer.Size mapSize) Build impot layer.
static Layer	buildOutputLayer(int classNum) Construct the output layer, the number of categories, determine the number of output units according to the number of categories.
static Layer	buildSampLayer(Layer.Size scaleSize) Build sampling layer.
double	getBias(int mapNo) Get Bias.
int	getClassNum() Get type.
double[][]	getError(int mapNo) Get the mapNo map of the residual.
double[][]	getError(int recordId, int mapNo) Get the residual of mapNo in the recordId record.
double[][][][]	getErrors() Get all the residuals (each record and each map).
double[][][][]	getKernel() Get kernel.
double[][]	getKernel(int i, int j) Get the convolution kernel of the previous layer i map to the current layer j map.
Layer.Size	getKernelSize() Get the size of the convolution kernel, only the kernel layer has kernelSize, and all other layers are null.
double[][]	getMap(int index) Get the index map matrix.
double[][]	getMap(int recordId, int mapNo) Set map.
double[][][][]	getMaps() »ñÈ¡batch¸÷¸ömap¾ØÕó
Layer.Size	getMapSize() Get map size.
int	getOutMapNum() Get number of maps.
Layer.Size	getScaleSize() Get sample size, only sampling layer has scaleSize, other layers are not null.
edu.hitsz.c102c.cnn.Layer.LayerType	getType() Get type.
void	initBias(int frontMapNum) ³õʼ»¯Æ«ÖÃ
void	initErros(int batchSize) Initiazlie errors.
void	initKernel(int frontMapNum) Subtract the size size and append a value append to x and y respectively.
void	initOutmaps(int batchSize) ³õʼ»¯Êä³ömap
void	initOutputKerkel(int frontMapNum, Layer.Size size) Randomly initialize the convolution kernel.
static void	prepareForNewBatch() Prepare for the next batch of training·
static void	prepareForNewRecord() Prepare the next recorded workout.
void	setBias(int mapNo, double value) Set bias.
void	setError(int mapNo, double[][] matrix)
void	setError(int mapNo, int mapX, int mapY, double value)
void	setKernel(int lastMapNo, int mapNo, double[][] kernel)
void	setMapSize(Layer.Size mapSize) Set map size.
void	setMapValue(int mapNo, double[][] outMatrix) ÒÔ¾ØÕóÐÎʽÉèÖõÚmapNo¸ömapµÄÖµ
void	setMapValue(int mapNo, int mapX, int mapY, double value) ÉèÖÃmapÖµ
void	setOutMapNum(int outMapNum) ÉèÖÃÊä³ömapµÄ¸öÊý

Methods inherited from class java.lang.Object

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

Method Detail

prepareForNewBatch

public static void prepareForNewBatch()

Prepare for the next batch of training·

prepareForNewRecord

public static void prepareForNewRecord()

Prepare the next recorded workout.

buildInputLayer

public static Layer buildInputLayer(Layer.Size mapSize)

Build impot layer.

Parameters:

mapSize -

Returns:

buildConvLayer

public static Layer buildConvLayer(int outMapNum,
                                   Layer.Size kernelSize)

Build convolution layer.

Returns:

buildSampLayer

public static Layer buildSampLayer(Layer.Size scaleSize)

Build sampling layer.

Parameters:

scaleSize -

Returns:

buildOutputLayer

public static Layer buildOutputLayer(int classNum)

Construct the output layer, the number of categories, determine the number of output units according to the number of categories.

Returns:

getMapSize

public Layer.Size getMapSize()

Get map size.

Returns:

setMapSize

public void setMapSize(Layer.Size mapSize)

Set map size.

Parameters:

mapSize -

getType

public edu.hitsz.c102c.cnn.Layer.LayerType getType()

Get type.

Returns:

getOutMapNum

public int getOutMapNum()

Get number of maps.

Returns:

setOutMapNum

public void setOutMapNum(int outMapNum)

ÉèÖÃÊä³ömapµÄ¸öÊý

Parameters:

outMapNum -

getKernelSize

public Layer.Size getKernelSize()

Get the size of the convolution kernel, only the kernel layer has kernelSize, and all other layers are null.

Returns:

getScaleSize

public Layer.Size getScaleSize()

Get sample size, only sampling layer has scaleSize, other layers are not null.

Returns:

initKernel

public void initKernel(int frontMapNum)

Subtract the size size and append a value append to x and y respectively.

Parameters:

frontMapNum -

initOutputKerkel

public void initOutputKerkel(int frontMapNum,
                             Layer.Size size)

Randomly initialize the convolution kernel.

Parameters:

frontMapNum -

size -

initBias

public void initBias(int frontMapNum)

³õʼ»¯Æ«ÖÃ

Parameters:

frontMapNum -

initOutmaps

public void initOutmaps(int batchSize)

³õʼ»¯Êä³ömap

Parameters:

batchSize -

setMapValue

public void setMapValue(int mapNo,
                        int mapX,
                        int mapY,
                        double value)

ÉèÖÃmapÖµ

Parameters:

mapNo - µÚ¼¸¸ömap

mapX - mapµÄ¸ß

mapY - mapµÄ¿í

value -

setMapValue

public void setMapValue(int mapNo,
                        double[][] outMatrix)

ÒÔ¾ØÕóÐÎʽÉèÖõÚmapNo¸ömapµÄÖµ

Parameters:

mapNo -

outMatrix -

getMap

public double[][] getMap(int index)

Get the index map matrix. In performance considerations, no duplicate objects are returned, but the reference is returned directly, and the caller must be careful. Avoid modifying outmaps. If you need to modify, please call setMapValue(...)

Parameters:

index -

Returns:

getKernel

public double[][] getKernel(int i,
                            int j)

Get the convolution kernel of the previous layer i map to the current layer j map.

Parameters:

i - ÉÏÒ»²ãµÄmapϱê

j - µ±Ç°²ãµÄmapϱê

Returns:

setError

public void setError(int mapNo,
                     int mapX,
                     int mapY,
                     double value)

Parameters:

mapNo -

mapX -

mapY -

value -

setError

public void setError(int mapNo,
                     double[][] matrix)

Parameters:

mapNo -

matrix -

getError

public double[][] getError(int mapNo)

Get the mapNo map of the residual. Does not return a copy of the object, but directly returns a reference, call the cautious, Avoid modifying errors. If you need to modify, please call setError(...)

Parameters:

mapNo -

Returns:

getErrors

public double[][][][] getErrors()

Get all the residuals (each record and each map).

Returns:

initErros

public void initErros(int batchSize)

Initiazlie errors.

Parameters:

batchSize -

setKernel

public void setKernel(int lastMapNo,
                      int mapNo,
                      double[][] kernel)

Parameters:

lastMapNo -

mapNo -

kernel -

getBias

public double getBias(int mapNo)

Get Bias.

Parameters:

mapNo -

Returns:

setBias

public void setBias(int mapNo,
                    double value)

Set bias.

Parameters:

mapNo -

value -

getMaps

public double[][][][] getMaps()

»ñÈ¡batch¸÷¸ömap¾ØÕó

Returns:

getError

public double[][] getError(int recordId,
                           int mapNo)

Get the residual of mapNo in the recordId record.

Parameters:

recordId -

mapNo -

Returns:

getMap

public double[][] getMap(int recordId,
                         int mapNo)

Set map.

Parameters:

recordId -

mapNo -

Returns:

getClassNum

public int getClassNum()

Get type.

Returns:

getKernel

public double[][][][] getKernel()

Get kernel.

Returns: