I have a very peculiar problem and I am not sure if SciDB can be the answer. I have a set of 128D feature vectors each having a unique identifier/name that I need to store in a database.
These points will be used for fast comparison to find a nearest neighbor using euclidean distance. Whenever my application extracts a new vector it needs to calculate the euclidean distance between the new vector and the vectors stored inside the db and return the closest vector as a match.
Is SciDB suitable for such operation? What do I need to make this work?