Optional Lab: Multiple linear regression (Initialization of w and b value)

Course Q&A Machine Learning Specialization Supervised ML: Regression and Classification
1

Initializing values of W and B
Initializing values of W and B1421×167 15.3 KB

Hi everyone!
I’m confused how the values of w and b in the optional lab are initialized. Can we select any random number for it? or is there a reason behind this selection? Thanks in advance!

2

It’s just an example, created by someone who already knew the best values.

In real work, the w and b values are learned through training.

For linear regression, typically the initial values will be all-zeros, because we don’t have any better guidance on the starting point.

3

Thank you :smile: @TMosh. Understood how the values are initialized.