Almost Navigable Graphs
For practitioners building large-scale vector databases, this work provides a theoretically grounded and practically effective way to construct sparser search graphs with competitive performance, addressing memory and speed bottlenecks of existing methods.
The paper introduces a relaxed notion of navigability called γ-almost navigability for approximate nearest neighbor search graphs, proving that any dataset admits such a graph with O(n/(1-γ)) edges (linear in dataset size) and a near-linear construction time. Empirically, these graphs achieve improved recall-runtime tradeoffs and less than half the degree of fully navigable graphs on billion-scale datasets.
Graph-based methods like HNSW, DiskANN, NSG, and others have become an increasingly popular choice for implementing approximate nearest neighbor search (ANNS) in Vector Databases (VecDBs). The success of these methods has motivated the study of how to best construct a search graph for a given dataset. To that end, \emph{navigability} has been identified as a desirable graph property which ensures good ANNS performance when combined with greedy search. However, for a dataset with $n$ vectors, the sparsest navigable graph requires $O(n\sqrt{n})$ edges in the worst-case, and we show empirically that, for typical billion node datasets, 100s of edges are needed per node. This leads to slow search and high memory requirements. Moreover, under standard complexity theoretical assumptions, it was recently established that constructing a sparse navigable graph requires $Ω(n^{2-ε})$ time, which is prohibitive for large datasets. We address these concerns by introducing a relaxed notation of navigability called ``$γ$-almost navigability'' for any $γ\in [0,1]$, with $γ= 1$ corresponding to full navigability. We prove that any dataset (under any distance) admits a $γ$-almost navigable graph with just $O\left(\frac{n}{1-γ}\right)$ edges, linear in the dataset size. We present a randomized algorithm for constructing such a graph in near-linear time. While we prove that $γ$-almost navigability sacrifices the worst-case search guarantees enjoyed by navigability, we show empirically that greedy beam search still performs well in such graphs when $γ< 1$. Indeed, we obtain improved recall-runtime tradeoffs on a variety of datasets compared to fully navigable graphs. Moreover, our graphs are more space efficient, with degree typically less than half that of a fully navigable graph for comparable performance.