#1




What the hint of Problem 3.6(b)
I have no idea about how to formulate the task of finding a separating w for separable data as a linear program. Could you probably tell me the first step of solving it?

#2




Re: What the hint of Problem 3.6(b)
Part (a) gives constraints that w must satisfy. These are the constraints in the linear program. Argue that you can choose c to be anything you want because any weights satisfying the constraints will work.
__________________
Have faith in probability 
#3




Re: What the hint of Problem 3.6(b)
Yes, I could understand this. The difficulty for me is to find the min and separate the optimization variable w. For Ein, it is hard to separate w in order to find c.

#4




Re: What the hint of Problem 3.6(b)
I don't see why you need to do "separate w to find c". If you satisfy the constraints, you have Ein=0 which is the minimum. So in this part, all you really need to do is satisfy the constraints, so you can literally choose c to be anything you want.
__________________
Have faith in probability 
Thread Tools  
Display Modes  

