I am traing a kNN with binary feature vectors. Why does it practically make no difference if I am using Euclidian distance or Hamming/Manhatten distance? Are they theoretically the same for binary data?
[link] [2 comments]
submitted by kumquatz
[link] [2 comments]
