“I highly recommend people give this paper a read.”
The Case for Learned Index Structures
主张用学习到的模型取代 B-Tree 等索引,最高提速 70%
论文把索引看作模型:B-Tree 是把键映射到有序数组中位置的模型,因此可以用包括深度学习在内的其他模型来替代,称为学习型索引。作者从理论上分析了学习型索引何时优于传统索引、设计中的主要挑战,初步结果显示在若干真实数据集上用神经网络可比缓存优化的 B-Tree 快最多 70%,同时内存占用小一个数量级。
HN 上的人把它当作用机器学习重新设计数据库和系统底层组件的代表想法,多数提及集中在 2017 至 2018 年,称它好读、有启发,并常以它为例说明神经网络可替代 B-Tree。
HN 书评
推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及“A fantastic paper on this from Jeff Dean at Google I suggest you read.”
“but the paper is an incredibly cool read and very approachable: The Case for Learned Index Structures”
“Thought one of the more interesting ideas is the Jeff Dean paper”
“Probably my favorite of late is from Jeff Dean”
“But my favorite new thing with these chips is the Jeff Dean paper.”
看全部 20 条 →
“My favorite paper right now is”
“Most interesting thing I have seen for for database indexes is this paper.”
“Paper is from Jeff Dean and a worth while read, IMO.”
“You should look at the paper they mention The Case for Learned Index Structures[0].”
“The topic of learned indexes is worth investigating”
“Under some assumptions it does handle inserts.”
“they replace B-Trees with neural nets:”
“speed up some classical algorithm based on heuristics by modeling the data with a neural network”
“The idea was you could train a model to predict the location of a row based on a primary key.”
“It would be something analogous to the 'learned indexes' paper.”
“Our initial results show, that by using neural nets we are able to outperform cache-optimized B-Trees by up to 70% in speed while saving an order-of-magnitude in memory over several real-world data sets.”
“The learned index[1] is basically a variation of this sort of”
“The Case for Learned Index Structures: The Case for Learned Index Structures”
“When NPU will become available in most systems, we will see learned indexes like in The Case for Learned Index Structures on client side.”