algorithm - Find two vectors whose sum is close to query vector -
algorithm - Find two vectors whose sum is close to query vector -
basically, question sort of research question. post here gain more hints you. suppose have set s of n d-dimension vectors. given d-dimension query vector, want find 2 vectors s sum close query vector, in term of cosine distance.
right now, solution enumerating possible combinations of s, results n^2 possible sum-vectors. hash these n^2 vectors b buckets using local sensitive hashing. when given query vector, hash query 1 of b buckets , check possible sum-vectors in bucket, see 1 closest. need o(b) space , o(n^2/b) searching time. space time trade off.
is there other solutions ? hints welcome ! don't need exact closest results, approximation results more welcome!
algorithm hash sum approximation
Comments
Post a Comment