Quote:
Originally Posted by Michael Reach
I've started working on the SVM part of the final. As I was browsing around online, I came across a piece of terminology that wasn't used in the lectures AFAIK, and I wondered what it meant.
The final asks, "The number of support vectors you get..."
But in various places I see descriptions of two types of support vectors: number of support vectors, and number of bounded support vectors. Does anyone know what is the difference between them, and which are the ones to count?

In this case, we are using hardmargin SVM and therefore there is only one type of support vectors (achieving the margin exactly). The different types arise in softmargin SVM because some support vectors there violate the margin.