Normalization of test data - neural network

i am using neural network for training and testing the data. i use features like nucleus area, perimeter, diameter, eccentricity for training.

i have values for features like:

Area - 2513, Perimeter - 203.811, Diameter - 56.5655, Eccentricity - 0.736247.

i have to normalize those values and have to save the normalized values in same '.mat' file to test the model. please help me to solve this, and provide me with the correct code.

Thank you in advance.

[Adam Danz]

Normalization could mean scaling values to a range of 0:1 but it could also have many other meanings [-1 : 1], [some min value : some max value], etc.

"please help me to solve this"

I'd be happy to help to steer you in the right direction but you should clear up what you mean by 'normalize'. Is your question more basic, such as "what is normalization"?

[vidhya v]

thank you for your response sir.

I'd be happy to help to steer you in the right direction but you should clear up what you mean by 'normalize'. Is your question more basic, such as "what is normalization"?

YES SIR. still i am not clear in this concept.

[Adam Danz]

I suggest you search for a tutorial to get you started in understanding normalization. I can provide one example that you're already familiar with: percentages.

Percentages convert a number from some absolute value to a values between 0 and 1 (or 0 and 100). If there are 7 people in a room with red hair and the room has 35 people, 20% (or 0.2) is the normalized value of people in the room with red hair.

To address your question, how to normalize the NN values, you'll need to dig a little deeper into your assignment and lecture notes to understand how your instructor defines normalization. There is not a single definition.

[vidhya v]

thank you sir

Answer 1

[Srivardhan Gadila]

The following resources might help you:

1. Normalization Help Center
2. normalize
3. In case of Multilayer Shallow Neural Networks you might use net.inputs{i}.processFcns
4. In case of Deep Neural Networks you might make use of 'Normalization' name-value pair argument for the Input Layers available in Deep Learning Toolbox

[vidhya v]

thank you sir.

and sorry for the delay