Abstract
An efficient procedure for finding Nearest Neighbours in metric spaces is the Approximating and Eliminating Search Algorithm (AESA). The expected number of distance computations required by the AESA is known to be asymptotically constant. However, the expected overall cost of the AESA, taking also into account computation not allotted to distance computation, is linear in the number of given points. A new variation of the AESA (ROAESA), in which the expected overall cost is sublinear, is introduced.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have