LFD Book Forum  

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

Reply
 
Thread Tools Display Modes
  #1  
Old 09-12-2012, 09:04 PM
munchkin munchkin is offline
Member
 
Join Date: Jul 2012
Posts: 38
Default Clarification Request On Problem 13

If a data set is not separable using RBF in the Z-space then the QP solver will hit the iteration limit and bomb. If the solver completes and returns an alpha vector that happens to provide zero in-sample error with the test data set then how does that indicate that the training data was not linearly separable? Is every occurrence of zero in-sample error during testing to be interpreted as a failure to separate the training data? I don't understand the operating procedure specified by the problem text.

Thanks for your attention.
Reply With Quote
  #2  
Old 09-12-2012, 09:27 PM
yaser's Avatar
yaser yaser is offline
Caltech
 
Join Date: Aug 2009
Location: Pasadena, California, USA
Posts: 1,477
Default Re: Clarification Request On Problem 14

Quote:
Originally Posted by munchkin View Post
If a data set is not separable using RBF in the Z-space then the QP solver will hit the iteration limit and bomb. If the solver completes and returns an alpha vector that happens to provide zero in-sample error with the test data set then how does that indicate that the training data was not linearly separable? Is every occurrence of zero in-sample error during testing to be interpreted as a failure to separate the training data? I don't understand the operating procedure specified by the problem text.

Thanks for your attention.
If you get \alpha's that achieve zero in-sample error, then the training data is separable by definition (those separating \alpha's are an existence proof of that fact).
__________________
Where everyone thinks alike, no one thinks very much
Reply With Quote
  #3  
Old 09-12-2012, 10:44 PM
munchkin munchkin is offline
Member
 
Join Date: Jul 2012
Posts: 38
Default Re: Clarification Request On Problem 14

Thanks for the prompt response.

Ein of zero is perfectly separated. I understand that. Perhaps my confusion arose from thinking about the PLA and how it can fail spectacularly when confronted with non-separable data. I had the impression that missing a few target function points during the in-sample error calculation wasn't a problem of the same magnitude as the solver being unable to converge to any solution at all.

I will talley up the Ein=0 statistics as specified in the problem instructions. Thanks.
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 04:35 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.