Practical Neural Network Recipies in C++

332

Chapter 18

void KohNet::in_norm (

double *input,

double *normfac ,

double *synth

)

{

double length, d ;

length = veclen ( nin , input ) ; // Squared length

if (normalization == 0) { // Multiplicative

*normfac = 1.0 / sqrt ( length ) ;

*synth = 0.0 ;

}

else if (normalization == 1) { // Z

*normfac =1.0/ sqrt ( nin ) ;

d = (double) nin - length ;

if (d > 0.0)

*synth = sqrt ( d ) * *normfac ;

else // If the inputs are all -1 to 1

*synth = 0.0 ; // this error never occurs

}

}

The variable nin is a KohNet class variable which is the number of

inputs (n in the above equations). Similarly, normalization is a class

variable that specifies which