1/31
Looks like no tags are added yet.

explain notation
L contains points on the line (v such that …
find min distance btw some point x and all points on the line → ends up being perpendicular distance

translate
x → classifier, magnitude or distance of x from origin
dividing by ||w||,
shortest distance btw x and line?

to proof this first
find upper bound and lower bound → becomes shortest distance

what does c tell you?
how many copies of w vector to subtract from x to be on line

how c vlaue derived


next step? and explain


explain
both u,v on line so


simplify this cross product


explain


finish showing lower bound





margin equation using xi, w, b

what are support vectors
closes training points on each side of the decision boundary
T/F: There can be more than one support vector on either side
true, every training point that lies on a margin line is a support vector
Hard-SVM: seek the separating hyperplane with the
largest margin
hard SVMf goal ormula


this is the same as


explain this


what is optimal soln to

what does this mean? margin is scale-sensitive:








prove



when data not separable, instead of hard-svm
use soft-svm
soft svm formula





explain why can we remove bias term


