Name
Abstract | ||
|---|---|---|
Permutation-based indexing is one of the most popular techniques for the approximate nearest-neighbor search problem in high-dimensional spaces. Due to the exponential increase of multimedia data, the time required to index this data has become a serious constraint. One of the possible steps towards faster index construction is utilization of massively parallel platforms such as the GPGPU architectures. In this paper, we have analyzed the computational costs of individual steps of the permutation-based index construction in a high-dimensional feature space and summarized our hybrid CPU-GPU solution. Our experience gained from this research may be utilized in other individual problems that require computing Lp distances in high-dimensional spaces, parallel top-k selection, or partial sorting of multiple smaller sets. We also provide guidelines how to balance workload in hybrid CPU-GPU systems. |
Year | DOI | Venue |
|---|---|---|
2017 | 10.1007/s11042-016-3677-7 | Multimedia Tools Appl. |
Keywords | Field | DocType |
GPU, Parallel, Permutation-based indexing, Approximate similarity search, Bitonic sorting | Partial sorting,Feature vector,Massively parallel,Workload,Computer science,Permutation,Search engine indexing,Theoretical computer science,General-purpose computing on graphics processing units,Search problem | Journal |
Volume | Issue | ISSN |
76 | 9 | 1573-7721 |
Citations | PageRank | References |
0 | 0.34 | 19 |
Authors | ||
3 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Martin Krulis | 1 | 76 | 13.27 |
| Hasmik Osipyan | 2 | 11 | 2.65 |
| Stéphane Marchand-Maillet | 3 | 1039 | 104.97 |