LFD Book Forum

LFD Book Forum (http://book.caltech.edu/bookforum/index.php)
-   Homework 2 (http://book.caltech.edu/bookforum/forumdisplay.php?f=131)
-   -   question on 2.8 and 2.9 (http://book.caltech.edu/bookforum/showthread.php?t=890)

yijun2011@yahoo.com 07-19-2012 02:42 PM

question on 2.8 and 2.9
 
Could someone please help to clarify:

Homework 2.8 says repeat the experiment 1000 times. Does it mean generate 1000 input training set +10% noise and find average E_in ? (as linear regression uses closed form solution for W, so there is nothing random if we run 1000 times on the same training set ).

Homework 2.9 says transform the training data set to a 6d feature space. If we have 1000 training sets created in 2.8, which one is 2.9 referring to?

Thanks a lot,

Yijun

samirbajaj 07-19-2012 03:06 PM

Re: question on 2.8 and 2.9
 
Quote:

Originally Posted by yijun2011@yahoo.com (Post 3544)
Could someone please help to clarify:

...(as linear regression uses closed form solution for W, so there is nothing random if we run 1000 times on the same training set ).

Correct, I generated 1000 new random points each time.

2.9 asks you about the weights.

-Samir

yaser 07-19-2012 03:07 PM

Re: question on 2.8 and 2.9
 
Quote:

Originally Posted by yijun2011@yahoo.com (Post 3544)
Could someone please help to clarify

In this, and in all problems where we repeat an experiment for a number of runs and average the results of these runs, all specifications pertain to an individual run. For example, generating the training set, noise, and target function, is for an individual run. You then get results from that run and repeat the entire process a number of times (1000 in this case).

The goal is to average out statistical fluctuations that occur from run to run, so that the final result is indicative of the general behavior, rather than the particulars of any one training set or one target function, etc.


All times are GMT -7. The time now is 10:03 PM.

Powered by vBulletin® Version 3.8.3
Copyright ©2000 - 2020, 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.