I have been reading Regularized Multi-Class Semi-Supervised Boosting and I can't seem to understand equation 2. From my understanding, the loss function "l" times the probability of "p" is the expected loss (i.e. how likely x is of class y). However, I have no idea what the "d" function is, the author never defines it.
Can someone please explain this equation to me? I've been trying to wrap my head around it.
[link][4 comments]
