decision boundary와 가장 가까운 입력 데이터들Note that αi\alpha_iαi will all be zero except for the support vectors because of Karush-Kuhn-Tucker Condition Σαy=0\Sigma \alpha y = 0Σαy=0so there will be at least one + and one - hw,b(x)=g(wTx+b)h_{w, b}(x) = g(w^Tx + b)hw,b(x)=g(wTx+b)