PLA (question 7): inital weight is too weak hint

[dobrokot]

If I take output of linear regression, I usually have a little (1-5 of 100) points which are misclassified. But it does not help PLA, because first addition Xi to W destabilize "almost good" value of W. I can even invert sign of L.R. output, it doesn't affect number of iterations (despite number of misclassified items on first iteration changes from 1-5 to 95-99, I verified)

But if I use (L.R. output)*N, it does help for PLA, greatly reduces number of iterations.

Is it supposed, that usage of linear regression output as W should greatly change number of iterations? If yes, does it means I should to search for other bugs in my code?

Output of L.R. should be used as is, or should be multiplied to make it "stronger" ?

[sanbt]

Originally Posted by dobrokot

If I take output of linear regression, I usually have a little (1-5 of 100) points which are misclassified. But it does not help PLA, because first addition Xi to W destabilize "almost good" value of W. I can even invert sign of L.R. output, it doesn't affect number of iterations (despite number of misclassified items on first iteration changes from 1-5 to 95-99, I verified)

But if I use (L.R. output)*N, it does help for PLA, greatly reduces number of iterations.

Is it supposed, that usage of linear regression output as W should greatly change number of iterations? If yes, does it means I should to search for other bugs in my code?

Output of L.R. should be used as is, or should be multiplied to make it "stronger" ?

I think for this question you should set N=10 and just use output of linear regression. It should reduce the number if iterations of PLA. I tried it for N = 100 and agree with you that linear regression doesn't help improve iterations.

[Anne Paulson]

This issue is discussed in another thread. Short answer: scaling the weights by N might make a difference, but for the homework question, we should follow the directions as given and not scale.

[Michael Reach]

Anne! Welcome back! (I was in your Data Analysis course.) Are you like a TA here or something? I know you took this course already.

[Anne Paulson]

Hi Michael!

My post was from the last iteration of the course (look at the date). I'm not taking it this time. I only happened to see your post because I was randomly surfing around. Good luck to all the DA students taking this course now.

[Michael Reach]

Oh, my bad. I did see your exalted Senior Member status and I thought that meant T.A. or such.