LFD Book Forum  

Go Back   LFD Book Forum > Course Discussions > Online LFD course > The Final

Reply
 
Thread Tools Display Modes
  #1  
Old 06-06-2012, 02:00 PM
ladybird2012 ladybird2012 is offline
Member
 
Join Date: Apr 2012
Posts: 32
Default Question 14 - bias term

For Question 15 we implement the RBF model, and there is a bias term to be calculated. It seems to me that there are two ways of going about this:

1) use a similar logic to what we had for linear regression. This seems to be what is suggested on slide 15 of lecture 16 (RBF network slide).

2) in the RBF notes it's mentioned in the footnote on page 27 that the bias = mean(y).

I implemented both and get different values for the bias as well as for Eout. I was wondering if anyone had tried both methods and come up with consistent values (i.e., I have a bug that I haven't managed to find despite looking hard and long) or if there is a reason to pick one method over the other.

Also while I'm at it, shouldn't the last line of the footnote only refer to w since the centers are chosen in an unsupervised way?
Reply With Quote
  #2  
Old 06-06-2012, 02:04 PM
yaser's Avatar
yaser yaser is offline
Caltech
 
Join Date: Aug 2009
Location: Pasadena, California, USA
Posts: 1,476
Default Re: Question 15 - bias term

Quote:
Originally Posted by ladybird2012 View Post
For Question 15 we implement the RBF model, and there is a bias term to be calculated. It seems to me that there are two ways of going about this:

1) use a similar logic to what we had for linear regression. This seems to be what is suggested on slide 15 of lecture 16 (RBF network slide).
This is the correct way. Treat b as you treated w_0 in linear regression. The only reason it's called b here is not to conflict with the SVM kernel version.
__________________
Where everyone thinks alike, no one thinks very much
Reply With Quote
  #3  
Old 09-05-2012, 01:22 PM
itooam itooam is offline
Senior Member
 
Join Date: Jul 2012
Posts: 100
Default Re: Question 15 - bias term

TTT (To The Top) Useful to know
Reply With Quote
  #4  
Old 09-17-2012, 01:51 AM
DavidNJ DavidNJ is offline
Member
 
Join Date: Jul 2012
Posts: 28
Default Re: Question 15 - bias term

What is the bias term in a linear regression? Aren't they solved as the pseudo inverse: inv(x'*x)*x'*y?
Reply With Quote
  #5  
Old 09-17-2012, 07:39 AM
yaser's Avatar
yaser yaser is offline
Caltech
 
Join Date: Aug 2009
Location: Pasadena, California, USA
Posts: 1,476
Default Re: Question 15 - bias term

Quote:
Originally Posted by DavidNJ View Post
What is the bias term in a linear regression? Aren't they solved as the pseudo inverse: inv(x'*x)*x'*y?
It is w_0 in our notation, corresponding to x_0=1 which is the first column of the matrix {\rm X} that is used in the pseudo-inverse.
__________________
Where everyone thinks alike, no one thinks very much
Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 10:26 PM.


Powered by vBulletin® Version 3.8.3
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
The contents of this forum are to be used ONLY by readers of the Learning From Data book by Yaser S. Abu-Mostafa, Malik Magdon-Ismail, and Hsuan-Tien Lin, and participants in the Learning From Data MOOC by Yaser S. Abu-Mostafa. No part of these contents is to be communicated or made accessible to ANY other person or entity.