About normalising data

Course Q&A Deep Learning Specialization Improving Deep Neural Networks: Hyperparameter tun
1

There seems to be a contradiction about how to normalise data precisely. After we X = X – µ, are we supposed to divide X by SD or variance?

In the “clarification about upcoming video” for normalisation in week 1, it says SD, but again in week 3, in the video about batch normalisation (around 1:10) it says variance. Somebody please clarify.

Also, how do we write math equations in this text field?

2

Hi, @BurntCalcium.

You should divide by the standard deviation: image (source)

I’ve opened an issue about this, thank you for taking the time to report it.

As of now, there’s no easy way to write math equations. You could use QuickLaTeX.com and paste the images here. Hopefully the Math Plugin will be enabled soon :slight_smile:

2 Likes