Almost Optimal Explicit Johnson-Lindenstrauss Transformations

Abstract

The Johnson-Lindenstrauss lemma is a fundamental result in probability with several applications in the design and analysis of algorithms in high dimensional geometry. Most known constructions of linear embeddings that satisfy the Johnson-Lindenstrauss property involve randomness. We address the question of explicitly constructing such embedding families and provide a construction with an almost optimal use of randomness: we use O(log(n/delta)log(log(n/delta)/epsilon)) random bits for embedding n dimensions to O(log(1/delta)/epsilon^2) dimensions with error probability at most delta, and distortion at most epsilon. In particular, for delta = 1/poly(n) and fixed epsilon, we use O(log n loglog n) random bits. Previous constructions required at least O(log^2 n) random bits to get polynomially small error.