Principal component analysis and higher correlations for distributed data
Abstract
We consider algorithmic problems in the setting in which the input data has been partitioned arbitrarily on many servers. The goal is to compute a function of all the data, and the bottleneck is the communication used by the algorithm. We present algorithms for two illustrative problems on massive data sets: (1) computing a low-rank approximation of a matrix A = A<sup>1</sup> + A<sup>2</sup> +... + A<sup>s</sup>, with matrix A<sup>t</sup> stored on server t and (2) computing a function of a vector a<inf>1</inf> + a<inf>2</inf> +... + a<inf>s</inf>, where server t has the vector a<inf>t</inf>; this includes the well-studied special case of computing frequency moments and separable functions, as well as higher-order correlations such as the number of subgraphs of a specified type occurring in a graph. For both problems we give algorithms with nearly optimal communication, and in particular the only dependence on n, the size of the data, is in the number of bits needed to represent indices and words (O(log n)).