VBLSH: Volume-balancing locality-sensitive hashing algorithm for K-nearest neighbors search

FLEX: A fast and light-weight learned index for kNN search in high-dimensional space

“…Given a d-dimensional dataset D with n points and a query point q, the kNN problem aims to find the top-k nearest neighbors of q, which is a subset R(q) ⊆ D, s.t., |R(q)| = k and ∀x ∈ R(q), ∀y ∈ D − R(q), dis(x, q) ≤ dis(y, q). Various techniques have been proposed for efficient kNN search, such as tree-based [2,3,4,7,14,24,37], LSH-based [11,12,13,33,35,38,39], quantization-based [18,19,27] approaches. However, retrieving the nearest neighbors from a large-scale and high-dimensional dataset is computationally challenging due to the curse of dimensionality, i.e., existing approaches either suffer from high construction cost or unsatisfactory search performance.…”