SDLOSS Compute loss based on confusion matrices

    L=SDLOSS(CM,COST)
    L=SDLOSS(CM,COST,PRIORS)

 INPUT
   CM     matrix CC x CC x OPC (CC=class count, OPC=op.point count)
   COST   matrix CC x CC
   PRIORS optional class priors (default: use equal priors)

 OUTPUT
   L      vector 1 x OPC of loss values (smaller is better)

 DESCRIPTION
 SDLOSS function computes loss based on confusion matrix.
 The lower the loss, the better.