LFD Book Forum

LFD Book Forum (http://book.caltech.edu/bookforum/index.php)
-   Homework 3 (http://book.caltech.edu/bookforum/forumdisplay.php?f=132)
-   -   *ANSWER* question 9 (http://book.caltech.edu/bookforum/showthread.php?t=4239)

Elroch 04-24-2013 12:57 AM

*ANSWER* question 9
 
I imagine that most people attacked question 9 in the way I did first of all, by guessing that points in a regular polygon configuration would be critical, and then seeing what is possible to achieve with a triangle.

I was wondering how many were a little unhappy with this and found the alternative route that I eventually chanced on. This involves embedding the 2 dimensional space in a 6-dimensional one, (by the map (x_1,x_2)\mapsto(x_1,x_2,x_1,x_2,x_1,x_2) and map each triangle hypothesis to a single perceptron hypothesis on the 6-dimensional space, by considering it as a combination of 3 perceptron hypotheses. This idea seems to have quite a lot of mileage for similar higher dimensional problems that would be intractable using ad hoc methods.

[EDIT: the amusing thing is that although this idea did give me more confidence in my answer, I can now see my mental reasoning was invalid, and this does not really justify the *ANSWER* label. However, I am now fairly happy with simple geometric reasoning based on visualisation that n-sided polygons in 2 dimensions have VC dimension 2n+1, and it seems rather a big co-incidence that this is the same as the VC dimension of perceptrons in \mathbb{R}^{2n} that my erroneous argument would give]

prithagupta.nsit 08-12-2015 03:04 PM

Re: *ANSWER* question 9
 
I see you have written for n sided polygon the VC dimension will be 2n+1 then why is the Vc dimension of rectangle is 4??


All times are GMT -7. The time now is 01:14 AM.

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.