Giters

skent259 / mildsvm

Multiple Instance Learning with Distributions, SVM

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Dual QP - find a better way to compute `b_`

skent259 opened this issue · comments

commented

In the kernel svm dual formation, the calculation of the intercept (b_) requires only using the alpha values that are in the interior of the $0 < \alpha_i < C$ constraint. The current implementation calculates b_ from alpha > 0, but I need to double the other side of this constraint and how the MI-SVM dual relates to that.