|
||
---|---|---|
LICENSE | ||
README.md | ||
cubes.go | ||
cubes_test.go | ||
go.mod | ||
hashes.go | ||
hashes_test.go |
README.md
Hashing float vectors in N-dimensions
This package allows fast approximate search of nearest neighbour vectors in n-dimensional space. It does not have any dependencies.
Its functions discretize a vector and generate a set of hashes, as described here (also as PDF).
Usage sequence:
- CubeSet or CentralCube, depending which one is used for a database record and which one for a query.
- HashSet and DecimalHash to get corresponding hash set and central hash from results of (2). If DecimalHash is not suitable because of very large number of buckets or dimensions, use FNV1aHash to get both the hash set and the central hash).
Example of usage for image comparison.
Go doc for code documentation.