Multilayer Perceptron Input Normalization

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Multilayer Perceptron Input Normalization

wessel van persie
Dear All,

Option C and I say:
C: Normalizing a numeric class will NOT be done. (Set this to not normalize the class if it's numeric). Default value: false
I: Normalizing the attributes will NOT be done. (Set this to not normalize the attributes). Default value: false

So if these values have their default values, Normalizing WILL be done.

How exactly are the input attributes normalized?
Using a z-transformation?
Or simply divide by mean value of the attribute?

Best regards,

Wessel

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Multilayer Perceptron Input Normalization

mhall
Administrator
On 11/10/10 12:54 AM, wessel van persie wrote:

> Dear All,
>
> Option C and I say:
> C: Normalizing a numeric class will NOT be done. (Set this to not
> normalize the class if it's numeric). Default value: false
> I: Normalizing the attributes will NOT be done. (Set this to not
> normalize the attributes). Default value: false
>
> So if these values have their default values, Normalizing WILL be done.
>
> How exactly are the input attributes normalized?
> Using a z-transformation?
> Or simply divide by mean value of the attribute?

(value - min) / (max - min).

Cheers,
Mark.

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Multilayer Perceptron Input Normalization

Kiran Deshmukh
This maybe too late now to reply as the original query was made several years
ago. However, this is for anyone who is looking for an answer to the
question of normalising attributes (and numeric class) in Multilayer
Perceptron of WEKA.

While WEKA normalises attributes to be between 0 and 1 while preprocessing
attributes, the Multilayer Perceptron normalises attributes and numeric
class values between -1 and +1. So, the formula is:

value_norm = 2*(value - min) / (max - min) -1

and:

value = (value_norm + 1)*(max - min) / 2 + min





--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html