When considering Support Vector Machine, in an take in multiple inputs. Can each of these inputs be a vector?? What i am trying to say is, can the input be a 2 dimensional vector??
$\begingroup$
$\endgroup$
1
-
$\begingroup$ It's not clear what you mean. Is the input to SVM a vector? yes, that's true for most ML algorithms. Multiple vectors? yes, training happens on > 1 data point. 2D? yes. Somehow, several 2D vectors at once? not clear what that means. Please clarify. $\endgroup$ – Sean Owen Jan 31 '15 at 11:32
Add a comment
|
$\begingroup$
$\endgroup$
1
If I understand your question correctly. Yes, SVM can take multiple inputs. My suggestion for handling a vector as a feature would be to expand it out. For example,
x0 = (1,2) x0 = 1
x1 = .4 -----> x1 = 2
x2 = 0 x2 = .4
x3 = 0
If this does not capture all of the characteristics of the vector that are important, then you may want to add other features (like magnitude of the vector) as well.
answered Jan 29 '15 at 12:42
-
$\begingroup$ ok, so if one of the input is 2d then it is expanded, yes, this is exactly what i wanted to know... thanks a lot $\endgroup$ – girl101 Jan 31 '15 at 9:56
