1

I want to know that data normalization that is required whether it must be applied to whole part of training set both input and output or input segment is enough.

saeed sheikholeslami
  • 98
  • 1
  • 8

1 Answers1

0

Whether you should normalize output depends on the type of neurons you use in your neural network and what type of output you expect. Find out the range of possible outputs for the given type of cell, and check that your target outputs falls within this range. If not, you will need to normalize.

The 'standard' neural network uses the sigmoid function which outputs a value between 0 and 1, so if the desired output doesn't fall in this range, you'll need to normalize.

Bernhard Barker
  • 54,589
  • 14
  • 104
  • 138
  • Yeah you mean i do try and error and there is no certain rule. – saeed sheikholeslami Feb 15 '13 at 09:03
  • @saeedsheikholeslami Not trial and error. Know the theory of the neural network or check the API for the target output range (which is likely 0-1). Analyse your desired output data to determine whether this is the same. – Bernhard Barker Feb 15 '13 at 10:55
  • this is my ann:(input layer is linear and neuron count is 1)x1 (hidden layer is sigmoid and neuron count is 25(aimless i had no idea))x2 (output layer is sigmoid and neuron count is 1) – saeed sheikholeslami Feb 15 '13 at 11:04
  • training set data simply is numbers between -10 to 10 associated with power 3 function. – saeed sheikholeslami Feb 15 '13 at 11:10
  • @saeedsheikholeslami When using sigmoid, if the desired output range is anything other than a boolean value or between 0 and 1, you need to normalize. [Here](http://www.faqs.org/faqs/ai-faq/neural-nets/part2/) is a resource that may help. – Bernhard Barker Feb 15 '13 at 11:47