closest / nearest matrices sorted from minimum distance threshold

Asked Aug 10 '16 at 08:05

Active Aug 10 '16 at 08:19

Viewed 48 times

I have a database with 33 000 000 matrices. My matrices are bitarrays, all with the size 5x32.(small matrices but large volume)

Then a get a new sample with a matrix with the same size, 5x32.

I want to find matrices in the db that are closest or in a minimum distance from the sample. (top100 or within a threshold)

I thought of iterate through all matrices in the db doing an XOR to find Hamming distance but that seems to be time-consuming.

Q1. Which method do you propose ranked from "easy to use" without need for super programming/math skills ? Would prefer anything in C, .NET or maybe CUDA.

Q2. Is there any benefit of using an Nvidia GPU and CUDA for this scenario ?

edited Aug 10 '16 at 08:19

asked Aug 10 '16 at 08:05

user5889512

This question is far too broad and I have voted to close it as such. – talonmies Aug 10 '16 at 08:17
Im not looking for a snippet of code but a direction to look in. The challenge is clearly stated. – user5889512 Aug 10 '16 at 16:46
Thank you m.s. Also found these that I would try to evaluate. https://stackoverflow.com/questions/32785803/similar-image-search-by-phash-distance-in-elasticsearch/33363007#33363007 – user5889512 Aug 11 '16 at 07:29
https://stackoverflow.com/questions/38277218/optimized-cuda-matrix-hamming-distance – user5889512 Aug 11 '16 at 07:30

closest / nearest matrices sorted from minimum distance threshold

0 Answers0