p-Norms, Precision and search time
The Reasonable and Foreseeable Effectivness Of the L2 Norm
Created on November 29|Last edited on November 29
Comment
Wall Time and Accuracy - 2 is the way 2 go
Counting Comparisons: L2 is still the best
There is no neat trend and everything other than L2 works well might be spurious.
Using numpy's L1, time seems to vary a bit less
But not super strong evidence either
Add a comment