MCGI adapts graph-based vector search to data manifolds via local intrinsic dimensionality to improve approximation quality and stability at billion scale.
The probability Psuccess of finding a neighbor in the improving cone corresponds to the ratio of the surface area of the spherical cap C(θmax) to the total surface area of Sd−1
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MCGI: Manifold-Consistent Graph Indexing for Billion-Scale Disk-Resident Vector Search
MCGI adapts graph-based vector search to data manifolds via local intrinsic dimensionality to improve approximation quality and stability at billion scale.