public class Network
extends java.lang.Object
Class representing a neural network
Modifier and Type | Field and Description |
---|---|
double |
a
Transfer function parameters
|
double[][] |
activation
Output of each node
|
double |
b_htan
Transfer function parameters
|
double |
b_log
Transfer function parameters
|
double[][] |
delta
Delta weights
|
double[][][] |
momentum
Momentum term
|
int[] |
Nhidden
Number of units in each layer
|
int |
Ninputs
Number of units in each layer
|
int |
Nlayers
Number of layers
|
int |
Noutputs
Number of units in each layer
|
java.lang.String[] |
transfer
Transfer function of each layer (LOG | HTAN | LINEAR)
|
double[][][] |
w
Matrix of weights
|
Constructor and Description |
---|
Network()
Empty Constructor.
|
Network(SetupParameters global)
Constructor
|
Modifier and Type | Method and Description |
---|---|
void |
GenerateOutput(double[] input)
Generate output at activation using input
|
void |
GenerateOutput(double[] input,
double[] output)
Generate output using input
|
void |
LoadNetwork(java.lang.String file_name)
Load network weights from a file
|
boolean |
NetClassifyPattern(double[] pattern)
Check if a pattern is correctly classified
|
int |
NetGetClassOfPattern(double[] pattern)
Return the class where a pattern is classified
|
void |
PrintWeights()
Print weights to screen.
|
void |
SaveNetwork(java.lang.String file_name,
boolean append)
Save network weights to a file
|
void |
SaveOutputFile(java.lang.String file_name,
double[][] data,
int n,
java.lang.String problem)
Save output data to a file
|
double |
TestNetworkInClassification(SetupParameters global,
double[][] data,
int npatterns)
Test network in classification
|
double |
TestNetworkInRegression(SetupParameters global,
double[][] data,
int npatterns)
Test network in regression.
|
void |
TrainNetwork(SetupParameters global,
double[][] data,
int npatterns)
Train network without cross validation
|
void |
TrainNetworkWithCrossvalidation(SetupParameters global,
Data data)
Train Network using cross validation
|
public int Nlayers
public int Ninputs
public int Noutputs
public int[] Nhidden
public double[][][] w
public double[][][] momentum
public double[][] delta
public double[][] activation
public final double a
public final double b_log
public final double b_htan
public java.lang.String[] transfer
public Network()
Empty Constructor.
public Network(SetupParameters global)
Constructor
global
- Global Definition parameterspublic void TrainNetworkWithCrossvalidation(SetupParameters global, Data data)
Train Network using cross validation
global
- Global Definition parametersdata
- Input datapublic void TrainNetwork(SetupParameters global, double[][] data, int npatterns)
Train network without cross validation
global
- Global Definition parametersdata
- Input datanpatterns
- Number of patternspublic double TestNetworkInClassification(SetupParameters global, double[][] data, int npatterns)
Test network in classification
global
- Global Definition parametersdata
- Input datanpatterns
- Number of patternspublic double TestNetworkInRegression(SetupParameters global, double[][] data, int npatterns)
Test network in regression.
global
- Global Definition parametersdata
- Input datanpatterns
- Number of patternspublic void GenerateOutput(double[] input)
Generate output at activation using input
input
- Input datapublic void GenerateOutput(double[] input, double[] output)
Generate output using input
input
- Input dataoutput
- Output datapublic void SaveNetwork(java.lang.String file_name, boolean append)
Save network weights to a file
file_name
- Output file nameappend
- Append or overwrite flagpublic void LoadNetwork(java.lang.String file_name)
Load network weights from a file
file_name
- Input file namepublic void PrintWeights()
Print weights to screen. Not used
public boolean NetClassifyPattern(double[] pattern)
Check if a pattern is correctly classified
pattern
- Pattern to checkpublic int NetGetClassOfPattern(double[] pattern)
Return the class where a pattern is classified
pattern
- Pattern to checkpublic void SaveOutputFile(java.lang.String file_name, double[][] data, int n, java.lang.String problem)
Save output data to a file
file_name
- Output file namedata
- Input datan
- Data matrix order (number of rows and columns)problem
- Type of problem (CLASSIFICATION | REGRESSION )