We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Information Retrieval

Title: Learned Sorted Table Search and Static Indexes in Small Model Space

Abstract: Machine Learning Techniques, properly combined with Data Structures, have resulted in Learned Static Indexes, innovative and powerful tools that speed-up Binary Search, with the use of additional space with respect to the table being searched into. Such space is devoted to the ML model. Although in their infancy, they are methodologically and practically important, due to the pervasiveness of Sorted Table Search procedures. In modern applications, model space is a key factor and, in fact, a major open question concerning this area is to assess to what extent one can enjoy the speed-up of Learned Indexes while using constant or nearly constant space models. We address it here by (a) introducing two new models, i.e., denoted {\bf KO-BFS} and {\bf SY-RMI}, respectively; (b) by systematically exploring, for the first time, the time-space trade-offs of a hierarchy of existing models, i.e., the ones in {\bf SOSD}, together with the new ones. We document a novel and rather complex time-space trade-off picture, which is very informative for users. We experimentally show that the {\bf KO-BFS }can speed up Interpolation Search and Uniform Binary Search in constant space. For other versions of Binary Search, our second model, together with the bi-criteria {\bf PGM} index, can achieve a speed-up with a model space of $0.05\%$ more than the one taken by the table, being competitive in terms of time-space trade-off with existing proposals. The {\bf SY-RMI} and the bi-criteria {\bf PGM} complement each other quite well across the various levels of the internal memory hierarchy. Finally, our findings are of interest to designers, since they highlight the need for further studies regarding the time-space relation in Learned Indexes.
Subjects: Information Retrieval (cs.IR); Databases (cs.DB); Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG)
ACM classes: E.1; I.2; H.2
Cite as: arXiv:2107.09480 [cs.IR]
  (or arXiv:2107.09480v5 [cs.IR] for this version)

Submission history

From: Giosué Lo Bosco [view email]
[v1] Mon, 19 Jul 2021 16:06:55 GMT (327kb,D)
[v2] Wed, 21 Jul 2021 13:56:52 GMT (630kb,D)
[v3] Fri, 23 Jul 2021 13:18:02 GMT (267kb,D)
[v4] Mon, 6 Dec 2021 07:41:28 GMT (745kb,D)
[v5] Tue, 7 Dec 2021 19:49:12 GMT (744kb,D)

Link back to: arXiv, form interface, contact.