TheCodeHere
TheCodeHere
Let me explain myself. Currently i'm working with the Large Margin Nearest Neighbor (LMNN), the algorithm needs to compute a priori the k nearest neighbors with the same class label...
Well, yes. But doing that only fix half of the problem. The function will still compute the distance of each sample from itself. I mean, i could look for the...
Technically you can make that assumption, but in some cases, like mine, i have some points that overlap and appear in first place, instead of the main point. So, the...