特征向量内存快速查找库 GitHub - spotify/annoy: Approximate Nearest Neighbors in C++/Python optimized for memory usage and loading/saving to disk

背景

Annoy

Annoy ( Approximate Nearest NeighborsOh Yeah) is a C++ library with Python bindings to search for points in space that are close to a given query point. It also creates large read-only file-based data structures that are mmappedinto memory so that many processes may share the same data.

Install

To install, simply do sudo pip install annoyto pull down the latest version from PyPI.

For the C++ version, just clone the repo and #include "annoylib.h".

Background

There are some other libraries to do nearest neighbor search. Annoy is almost as fast as the fastest libraries, (see below), but there is actually another feature that really sets Annoy apart: it has the ability to use static files as indexes. In particular, this means you can share index across processes. Annoy also decouples creating indexes from loading them, so you can pass around indexes as files and map them into memory quickly. Another nice thing of Annoy is that it tries to minimize memory footprint so the indexes are quite small.

Why is this useful? If you want to find nearest neighbors and you have many CPU's, you only need the RAM to fit the index once. You can also pass around and distribute static files to use in production environment, in Hadoop jobs, etc. Any process will be able to load (mmap) the index into memory and will be able to do lookups immediately.

We use it at Spotifyfor music recommendations. After running matrix factorization algorithms, every user/item can be represented as a vector in f-dimensional space. This library helps us search for similar users/items. We have many millions of tracks in a high-dimensional space, so memory usage is a prime concern.

Annoy was built by Erik Bernhardssonin a couple of afternoons during Hack Week.

Summary of features

• Euclidean distance, Manhattan distance, cosine distance, Hamming distance, or Dot (Inner) Product distance
• Cosine distance is equivalent to Euclidean distance of normalized vectors = sqrt(2-2*cos(u, v))
• Works better if you don't have too many dimensions (like <100) but seems to perform surprisingly well even up to 1,000 dimensions
• Small memory usage
• Lets you share memory between multiple processes
• Index creation is separate from lookup (in particular you can not add more items once the tree has been created)
• Native Python support, tested with 2.6, 2.7, 3.3, 3.4, 3.5

Python code example

fromannoyimportAnnoyIndeximportrandom

f=40t=AnnoyIndex(f)#Length of item vector that will be indexedforiinxrange(1000):
    v=[random.gauss(0,1)forzinxrange(f)]
    t.add_item(i, v)

t.build(10)#10 treest.save('test.ann')#...u=AnnoyIndex(f)
u.load('test.ann')#super fast, will just mmap the fileprint(u.get_nns_by_item(0,1000))#will find the 1000 nearest neighbors

Right now it only accepts integers as identifiers for items. Note that it will allocate memory for max(id)+1 items because it assumes your items are numbered 0 … n-1. If you need other id's, you will have to keep track of a map yourself.

Full Python API

• AnnoyIndex(f, metric='angular')returns a new index that's read-write and stores vector of fdimensions. Metric can be "angular", "euclidean", "manhattan", "hamming", or "dot".
• a.add_item(i, v)adds item i(any nonnegative integer) with vector v. Note that it will allocate memory for max(i)+1items.
• a.build(n_trees)builds a forest of n_treestrees. More trees gives higher precision when querying. After calling build, no more items can be added.
• a.save(fn)saves the index to disk.
• a.load(fn)loads (mmaps) an index from disk.
• a.unload()unloads.
• a.get_nns_by_item(i, n, search_k=-1, include_distances=False)returns the nclosest items. During the query it will inspect up to search_knodes which defaults to n_trees * nif not provided. search_kgives you a run-time tradeoff between better accuracy and speed. If you set include_distancesto True, it will return a 2 element tuple with two lists in it: the second one containing all corresponding distances.
• a.get_nns_by_vector(v, n, search_k=-1, include_distances=False)same but query by vector v.
• a.get_item_vector(i)returns the vector for item ithat was previously added.
• a.get_distance(i, j)returns the distance between items iand j. NOTE: this used to return the squareddistance, but has been changed as of Aug 2016.
• a.get_n_items()returns the number of items in the index.

Notes:

• There's no bounds checking performed on the values so be careful.
• Annoy uses Euclidean distance of normalized vectors for its angular distance, which for two vectors u,v is equal to sqrt(2(1-cos(u,v)))

The C++ API is very similar: just #include "annoylib.h"to get access to it.

Tradeoffs

There are just two parameters you can use to tune Annoy: the number of trees n_treesand the number of nodes to inspect during searching search_k.

• n_treesis provided during build time and affects the build time and the index size. A larger value will give more accurate results, but larger indexes.
• search_kis provided in runtime and affects the search performance. A larger value will give more accurate results, but will take longer time to return.

If search_kis not provided, it will default to n * n_trees * Dwhere nis the number of approximate nearest neighbors and Dis a constant depending on the metric. Otherwise, search_kand n_treesare roughly independent, i.e. a the value of n_treeswill not affect search time if search_kis held constant and vice versa. Basically it's recommended to set n_treesas large as possible given the amount of memory you can afford, and it's recommended to set search_kas large as possible given the time constraints you have for the queries.

How does it work

Using random projectionsand by building up a tree. At every intermediate node in the tree, a random hyperplane is chosen, which divides the space into two subspaces. This hyperplane is chosen by sampling two points from the subset and taking the hyperplane equidistant from them.

We do this k times so that we get a forest of trees. k has to be tuned to your need, by looking at what tradeoff you have between precision and performance.

Hamming distance (contributed by Martin Aumüller) packs the data into 64-bit integers under the hood and uses built-in bit count primitives so it could be quite fast. All splits are axis-aligned.

Dot Product distance (contributed by Peter Sobot) reduces the provided vectors from dot (or "inner-product") space to a more query-friendly cosine space using a method by Bachrach et al., at Microsoft Research, published in 2014.

More info

• Dirk Eddelbuettelprovides an R version of Annoy.
• Andy Sloaneprovides a Java version of Annoyalthough currently limited to cosine and read-only.
• Pishen Tsaiprovides a Scala wrapper of Annoywhich uses JNA to call the C++ library of Annoy.
• There is experimental support for Goprovided by Taneli Leppä.
• Boris Nagaevwrote Lua bindings.
• During part of Spotify Hack Week 2016 (and a bit afterward), Jim Kangwrote Node bindingsfor Annoy.
• Min-Seok Kimbuilt a Scala versionof Annoy.
• Presentation from New York Machine Learning meetupabout Annoy
• Radim Řehůřek's blog posts comparing Annoy to a couple of other similar Python libraries: Intro, Contestants, Querying
• ann-benchmarksis a benchmark for several approximate nearest neighbor libraries. Annoy seems to be fairly competitive, especially at higher precisions:

Source code

It's all written in C++ with a handful of ugly optimizations for performance and memory usage. You have been warned :)

The code should support Windows, thanks to Qiang Kouand Timothy Riley.

To run the tests, execute python setup.py nosetests. The test suite includes a big real world dataset that is downloaded from the internet, so it will take a few minutes to execute.

Discuss

Feel free to post any questions or comments to the annoy-usergroup. I'm @fulhackon Twitter.