arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1702.05760cs.DScs.CCcs.CGcs.CR
收藏

近似近邻的超立方体LSH

Hypercube LSH for approximate near neighbors

Thijs Laarhoven

摘要:

一个著名的寻找角距离近邻的技术涉及使用一组\textit{random}超平面将空间划分为散列区域[Charikar,STOC 2002]。后来的实验表明,使用一组\textit{orthogonal}超平面,从而将空间划分为由超立方体诱导的Voronoi区域,导致甚至更好的结果[Terasawa和Tanaka,WADS 2007]。然而,对于这种改进没有给出任何理论解释,并且仍然不清楚由此产生的超立方体哈希方法如何在高维中扩展。在这项工作中,我们提供了明确的渐近的碰撞概率时,使用超立方体划分的空间。例如,与使用随机超平面时的$(\frac{1}{2})^d$相比,预期两个接近正交的向量在维度$d$中以概率$(\frac{1}π)^{d + o(d)}$冲突。与随机超平面的$(\frac{2}{3})^d$相比,角度为$\fracπ{3}$的向量以概率$(\frac{\sqrt{3}}π)^{d + o(d)}$碰撞,并且在两种情况下,接近平行的向量以相似的渐近概率碰撞。对于$c$ -近似最近邻搜索,这转化为与超平面LSH相比,局部敏感散列(LSH)方法的指数$ρ$降低了高达$\log_2(π) \approx 1.652$的因子。对于$c = 2$,我们得到了超立方体LSH的$ρ\approx 0.302 + o(1)$,改进了超平面LSH的$ρ\approx 0.377$.我们进一步描述了如何在实践中使用超立方体LSH,我们认为在格算法领域的一个例子中的应用。

英文摘要:

A celebrated technique for finding near neighbors for the angular distance involves using a set of \textit{random} hyperplanes to partition the space into hash regions [Charikar, STOC 2002]. Experiments later showed that using a set of \textit{orthogonal} hyperplanes, thereby partitioning the space into the Voronoi regions induced by a hypercube, leads to even better results [Terasawa and Tanaka, WADS 2007]. However, no theoretical explanation for this improvement was ever given, and it remained unclear how the resulting hypercube hash method scales in high dimensions. In this work, we provide explicit asymptotics for the collision probabilities when using hypercubes to partition the space. For instance, two near-orthogonal vectors are expected to collide with probability $(\frac{1}π)^{d + o(d)}$ in dimension $d$, compared to $(\frac{1}{2})^d$ when using random hyperplanes. Vectors at angle $\fracπ{3}$ collide with probability $(\frac{\sqrt{3}}π)^{d + o(d)}$, compared to $(\frac{2}{3})^d$ for random hyperplanes, and near-parallel vectors collide with similar asymptotic probabilities in both cases. For $c$-approximate nearest neighbor searching, this translates to a decrease in the exponent $ρ$ of locality-sensitive hashing (LSH) methods of a factor up to $\log_2(π) \approx 1.652$ compared to hyperplane LSH. For $c = 2$, we obtain $ρ\approx 0.302 + o(1)$ for hypercube LSH, improving upon the $ρ\approx 0.377$ for hyperplane LSH. We further describe how to use hypercube LSH in practice, and we consider an example application in the area of lattice algorithms.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑