I A.2.2 Multi-class AMNs For K agt; 2, we use the randomized rounding procedure of Kleinberg and Tardos  to produce an integer solution for the linear relaxation, losing at most a factor of m = maxcec \c\ in the objective function. The basicanbsp;...
|Title||:||Learning structured prediction models|
|Author||:||Ben Taskar, Stanford University. Computer Science Dept|