Name
Abstract | ||
|---|---|---|
We consider the problem of encoding a finite set of vectors into a small number of bits while approximately retaining information on the angular distances between the vectors. By deriving improved variance bounds related to binary Gaussian circulant embeddings, we largely fix a gap in the proof of the best known fast binary embedding method. Our bounds also show that well-spreadness assumptions on the data vectors, which were needed in earlier work on variance bounds, are unnecessary. In addition, we propose a new binary embedding with a faster running time on sparse data. |
Year | DOI | Venue |
|---|---|---|
2018 | 10.1007/s00454-017-9964-x | Discrete and Computational Geometry |
Keywords | DocType | Volume |
Binary embeddings, Johnson–Lindenstrauss embeddings, Circulant matrices, 60B20, 68Q87 | Journal | abs/1608.06498 |
Issue | ISSN | Citations |
3 | 0179-5376 | 2 |
PageRank | References | Authors |
0.36 | 14 | 2 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Sjoerd Dirksen | 1 | 32 | 2.75 |
| Alexander Stollenwerk | 2 | 2 | 0.70 |