论文标题
Falconn ++：一种局部敏感的过滤方法，用于大约最近的邻居搜索
Falconn++: A Locality-sensitive Filtering Approach for Approximate Nearest Neighbor Search
论文作者
论文摘要
我们提出了Falconn ++，这是一种新型局部敏感的过滤方法，可在角度距离上进行大约最近的邻居搜索。 Falconn ++可以在任何哈希桶\ textit {}查询中滤除潜在的遥远点，与其他基于哈希的解决方案相比，这会导致更高质量的候选者。从理论上讲，Falconn ++渐近地达到的查询时间复杂性比Falconn（一种在角度距离上是对位置敏感的散列方案）的较低。从经验上讲，Falconn ++在许多现实世界中的数据集中都比Falconn实现更高的召回速度权衡。 Falconn ++也与HNSW竞争，HNSW是高度搜索召回机制的有效代表。
We present Falconn++, a novel locality-sensitive filtering approach for approximate nearest neighbor search on angular distance. Falconn++ can filter out potential far away points in any hash bucket \textit{before} querying, which results in higher quality candidates compared to other hashing-based solutions. Theoretically, Falconn++ asymptotically achieves lower query time complexity than Falconn, an optimal locality-sensitive hashing scheme on angular distance. Empirically, Falconn++ achieves higher recall-speed tradeoffs than Falconn on many real-world data sets. Falconn++ is also competitive with HNSW, an efficient representative of graph-based solutions on high search recall regimes.
