# SVM Dual Formulation :: KKT Constraint

In Andew Ng's SVM course notes, the final hard margin optimization problem is given as the following:

I am unclear how to see from this where the 5th constraint is satisfied.

The definition of $g_{i}$ is as follows:

Can anyone help explain?

• What does $g_i$ represent? – Marc Claesen Nov 21 '14 at 8:22
• The full lecture notes are here: cs229.stanford.edu/notes/cs229-notes3.pdf – B_Miner Nov 21 '14 at 13:50
• I added above specifically what is $g_{i}$. – B_Miner Nov 21 '14 at 13:54
• I am thinking this constraint (5) that either $a_{i}$ or $g_{i}$ must be zero, has to do with the Lagrangian dual but what I am not seeing is how that constraint is embodied in the optimization problem he has written. – B_Miner Nov 21 '14 at 14:33