*ANSWER* HW2 question 8

[dlammerts]

Question 8: Plot of non-linear training set (without and with noise added) used in HW2 #8 as a hint for anyone having difficulties visualizing why a linear classifier will have an in-sample error of approximately 0.5.

http://www.venturephilosopher.com/?page_id=109

Dirk

[yaser]

Originally Posted by dlammerts

Question 8: Plot of non-linear training set (without and with noise added) used in HW2 #8 as a hint for anyone having difficulties visualizing why a linear classifier will have an in-sample error of approximately 0.5.

http://www.venturephilosopher.com/?page_id=109

Dirk

Nice!

Here are the two figures from the linked page; the noiseless version:



and the noisy version:

[wangkexue]

because the function is a circle