
#1




PLA Optimization criteria
The PLA algorithm, eqn. 1.3, can be used to partition linearly separable data. What I'm curious is to what optimization criteria underlies eqn. 1.3? The figures on pp. 67 show that for a 2D case we have the algorithm converge to some straight line decision boundary, and it is also qualitatively clear that many different straightlines, would "work" equally well (ie give the same E_{in} error rate); however PLA converges to a specific solution. The PLA algorithm seems to provide both, an optimization criteria, and a method for solution too. The opt. criteria gives provides uniqueness. Can the optimization criteria underlying PLA (eqn 1.3) be spelled out explicitly? Thank you.
__________________
The whole is simpler than the sum of its parts.  Gibbs 
#2




Re: PLA Optimization criteria
Quote:
__________________
Where everyone thinks alike, no one thinks very much 
#3




Re: PLA Optimization criteria
A few weeks ago I recall having read that section on SGD, however the connection with PLA somehow slipped past. My sincere apologies. Then, I suppose I was trying keep my focus on ML paradigms & approaches, and much as optimization is part & parcel of ML, I think I tried not to get sidetracked with finer details of optimization. Lately, I've started rereading the book, a bit more carefully, and find myself appreciating the whole, and the subtle, even more so than before! Thank you for taking the trouble of pointing out the section. I do appreciate it.
__________________
The whole is simpler than the sum of its parts.  Gibbs 
Thread Tools  
Display Modes  

