CODEWITHSUNDEEP
machine-learningneural-networkdatasetconv-neural-networkfeed-forward
Rudina
Rudina

Reputation: 23

Transforming data in Feed-forward model in machine learning

I'm building a prediction model using feed-forward machine learning algorithm, and I read a lot about preprocessing training data (transforming data) by ranging it between 0 and 1 to improve learning. My question is: should I transform both the training samples(input) and labels(output) of the data or only the data samples?

Upvotes: -2

Views: 56

Answers (1)

GrantorShadow
GrantorShadow

Reputation: 41

You should only preprocess the training data and if needed the test data or real life data which will be used for generating inference from the model. You are only normalizing data by transforming it in between 0 & 1. This is done to get the data values on the same scale (homogeneity). Also, keep the variances same on both training and test data

TL;DR: Just normalize the training samples and the test data NOT the labels

Upvotes: 0

Related Questions