Regarding approximate hamming nearest neighbor implementation

[ghost]

https://github.com/jakevdp/klsh/blob/master/klsh/hamming_ann.py#L163 I have a doubt regarding your implementation of HammingANN. Your are not permuting the query bit vectors for HammingANN but you are permuting your stored data bit vectors. As per my understanding of the paper http://www.cs.princeton.edu/courses/archive/spr04/cos598B/bib/CharikarEstim.pdf query and stored data bit vectors should be permuted and binary search is done for each of the permutation to find the approximate nearest neighbor.

[jakevdp]

Thanks – this implementation is very much a work in progress, but one on which I'm not progressing much lately. I'm aware that it doesn't follow the paper exactly at the moment. If you'd like to contribute a PR with these changes, that would be great!