Approximate nearest neighbor search method combining VP tree and guide nearest neighbor graph

Abstract

The invention provides an approximate nearest neighbor search method combining a VP tree and a guide nearest neighbor graph. The method comprises two processes: an index construction process and a search process. The index construction process comprises the following steps: (1) constructing a VP tree for storing entry points for high-dimensional vector data; (2) constructing the high-dimensional vector data into a K-nearest neighbor graph by using a navigable extension graph; and (3) uniformly dividing the neighbor set of each point in the K neighbor graph to obtain a steerable neighbor graph.The search process comprises the following steps of: (1) searching an entry point, namely quickly obtaining the entry point approximately close to a query point by searching the VP tree; (2) a guidesearch stage: starting from the entry point, obtaining a local nearest point of a query point; and (3) an exhaustive search stage: exhaustively searching points near the query point by utilizing a range search algorithm and returning top-K nearest points of the query point. According to the method, the performance of approximate nearest neighbor search is effectively improved by using a hybrid search strategy.

Description

technical field [0001] The invention relates to the field of approximate nearest neighbor search, in particular to an approximate nearest neighbor search method combining VP tree and guided neighbor graph. Background technique [0002] Approximate nearest neighbor search is a key technique in data mining, machine learning, machine vision, and information retrieval, etc. In the past decades, people have been working on improving the performance of approximate nearest neighbor search. At present, the approximate nearest neighbor search includes the following four types of methods: tree-based, hash-based, quantization-based, and neighbor graph-based methods. Compared with other methods, the neighbor graph-based method is characterized by higher accuracy and faster speed. , has become a research hotspot in recent years. [0003] At present, excellent methods such as Hierarchical Navigable Small World Graph (HNSW), Navigable Satellite System Graph (NSSG), and Navigable Stretch G...

AI Technical Summary

Problems solved by technology

[0003] At present, excellent methods such as Hierarchical Navigable Small World Graph (HNSW), Navigable Satellite System Graph (NSSG), and Navigable Stretch Graph (NSG) have been proposed, and they have been widely used in actual production environments. However, There is a problem in most of the above-mentioned approximate nearest neighbor search methods based on the nearest neighbor graph: the graph search strategy of the method based on the nearest neighbor graph only uses a single greedy search algorithm, which will lead to low efficiency of the search, because the greedy search algorithm In the search process, it is necessary to traverse all the neighbor points of the current access point and perform similarity calculation with the query point, which is completely unnecessary in the search area far from the query point, and will seriously reduce the search efficiency, especially for large-scale high-dimensional In the case of data, the optimization of search efficiency is a major challenge

Images