public class RpropParameters
extends java.lang.Object
Constructor and Description |
---|
RpropParameters()
Creates a new instance of RpropParameters
|
Modifier and Type | Method and Description |
---|---|
int |
getBatchSize()
Gets the batchsize.
|
double |
getEtaDec()
Gets the decremental learning factor/rate.
|
double |
getEtaInc()
Gets the incremental learning factor/rate.
|
double |
getInitialDelta(int i,
int j)
Gets the initial delta value.
|
double |
getMaxDelta()
Gets the maximum allowed delta value.
|
double |
getMinDelta()
Gets the minimum allowed delta value.
|
void |
setBatchSize(int aBatchsize)
Sets the batchsize.
|
void |
setEtaDec(double anEtaDec)
Sets the decremental learning factor/rate.
|
void |
setEtaInc(double anEtaInc)
Sets the incremental learning factor/rate.
|
void |
setInitialDelta(double anInitialDelta)
Sets the initial delta for all delta's.
|
void |
setMaxDelta(double aMaxDelta)
Sets the maximum allowed delta value.
|
void |
setMinDelta(double aMinDelta)
Sets the minimum allowed delta value.
|
public RpropParameters()
public double getInitialDelta(int i, int j)
i
- the index (i, j) of the weight/bias for which it should get the
initial value. The RPROP learning algorithm gives every bias/weight
the same initial value, but by passing the index of the weight/bias
to this method, a user is able to give different initial values to
different weights/biases based on their index by extending this
class.j
- public void setInitialDelta(double anInitialDelta)
anInitialDelta
- the initial delta value.public double getMaxDelta()
public void setMaxDelta(double aMaxDelta)
aMaxDelta
- the maximum allowed delta value.public double getMinDelta()
public void setMinDelta(double aMinDelta)
aMinDelta
- the minimum allowed delta value.public double getEtaInc()
public void setEtaInc(double anEtaInc)
anEtaInc
- the incremental learning factor/rate.public double getEtaDec()
public void setEtaDec(double anEtaDec)
anEtaDec
- the decremental learning factor/rate.public int getBatchSize()
public void setBatchSize(int aBatchsize)
Submit Feedback to pmarrone@users.sourceforge.net