Locality Sensitive Hashing Part 2, Jeffry D Ullman

[@kemot25]

I think there is an error. At 42:28 he says "the more you raise it (pause) the more false negative you get" of course should be less FN you get. Moreover to decrease FP we should take less than 1024 not more. While the FN is growing the FP is dropping, at least here. Can someone support me or correct me?

[@MattMartin0]

Yes, you are right. The formula is 1 - x^y where x is a fraction between 0 and 1. He suggests making y bigger which, since x is less than 1, makes the resulting value larger. Though in the end he seems to suggest that increasing the number of points was what really mattered here even though he doesn't get into the math behind that assertion.

[@TMAC02010]

great!