Name

ANN_NORM — Function to normalise data to train a feed forward network

Calling Sequence

[DATA_final,DATA_mean,DATA_std] = ANN_NORM(DATA,[OBJ,MULT,MEAN,STD])

Parameters

DATA

Input data (matrix [PxN] where P is the number of input neurons and N the number of input patterns)

OBJ

Objective of the normalisation. 'raw2norm' = from raw values to normalised ones (default), 'norm2raw' = from normalised values to raw ones.

MULT

Multiplicator of standard deviation to use for the normalisation (default = 1). MULT can be a single value or a [Px1] vector.

MEAN

Means to use in the normalisation process (matrix [Px1])

STD

Standard deviations to use in the normalisation process (matrix [Px1])

DATA_final

DATA transformed by the normaliation of the denormalisation process

DATA_mean

Means used to normalise DATA (vector [Px1])

DATA_std

Standard deviations used to normalise DATA (vector [Px1])

Description

  • Add here a paragraph of the function description. Other paragraph can be added

  • Add here a paragraph of the function description

Examples

    // Raw values
    INraw     = 200*rand(6,100)-50;
    TARGraw   = 1000*rand(1,100)-888;   

   // Calibration of a network with 6 input nodes, 4 nodes in the hidden layer and 1 output node
   IN         = ANN_NORM(INraw);
   [TARG,m,s] = ANN_NORM(TARGraw);
   Wini = rand(4,7,2);
   [W,OUT,RMSE] = ANN_LMBR(IN,TARG,4,Wini,'def','def',%f); // %f => plot not shown

   // Plot results
   xset('window',0);
   plot([TARGraw' ANN_NORM(OUT','norm2raw',1,m,s)]); 

  

See Also

ANN_CONV_W , ANN_JACOB , ANN_LMBR , ANN_SIM

Authors

Julien Lerat

CEMAGREF Antony, HBAN Unit, julien.lerat@cemagref.fr