Journal
MATHEMATICS OF OPERATIONS RESEARCH
Volume 43, Issue 4, Pages 1051-1071Publisher
INFORMS
DOI: 10.1287/moor.2017.0894
Keywords
Johnson-Lindenstrauss lemma; concentration of measure; dimension reduction
Funding
- SO-grid project - ADEME
- EU [316647]
- ANR [ANR-10-BINF-0003]
- Microsoft Research PhD fellowship
Ask authors/readers for more resources
Random projections are random linear maps, sampled from appropriate distributions, which approximately preserve certain geometrical invariants so that the approximation improves as the dimension of the space grows. The well known Johnson-Lindenstrauss lemma states that there are random matrices with surprisingly few rows which approximately preserve pairwise Euclidean distances among a set of points. This is commonly used to speed up algorithms based on Euclidean distances. We prove that these matrices also preserve other quantities, such as the distance to a cone. We exploit this result to devise a probabilistic algorithm to approximately solve linear programs. We show that this algorithm can approximately solve very large randomly generated LP instances. We also showcase its application to an error correction coding problem.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available