For-All Sparse Recovery in Near-Optimal Time

An approximate sparse recovery system in l1 norm consists of parameters k, e, N; an m-by-N measurement F; and a recovery algorithm R. Given a vector, x, the system approximates x by &xwidehat; = R(F x), which must satisfy || &xwidehat;-x||1 (1+e)|| x - xk||1. We consider the "for all" model, in which a single matrix F, possibly "constructed" non-explicitly using the probabilistic method, is used for all signals x. The best existing sublinear algorithm by Porat and Strauss [2012] uses O(e3klog (N/k)) measurements and runs in time O(k1 aNa) for any constant a textgreater 0. In this article, we improve the number of measurements to O(e 2klog (N/k)), matching the best existing upper bound (attained by super-linear algorithms), and the runtime to O(k1+vpoly(log N,1/e)), with a modest restriction that k N1 a and e (log k/log N)g for any constants a, v, g textgreater 0. When k log cN for some c textgreater 0, the runtime is reduced to O(kpoly(N,1/e)). With no restrictions on e, we have an approximation recovery system with m = O(k/elog (N/k)((log N/log k)g + 1/e)) measurements. The overall architecture of this algorithm is similar to that of Porat and Strauss [2012] in that we repeatedly use a weak recovery system (with varying parameters) to obtain a top-level recovery algorithm. The weak recovery system consists of a two-layer hashing procedure (or with two unbalanced expanders for a deterministic algorithm). The algorithmic innovation is a novel encoding procedure that is reminiscent of network coding and that reflects the structure of the hashing stages. The idea is to encode the signal position index i by associating it with a unique message mi, which will be encoded to a longer message m'i (in contrast to Porat and Strauss [2012] in which the encoding is simply the identity). Portions of the message m'i correspond to repetitions of the hashing, and we use a regular expander graph to encode the linkages among these portions. The decoding or recovery algorithm consists of recovering the portions of the longer messages m'i and then decoding to the original messages mi, all the while ensuring that corruptions can be detected and/or corrected. The recovery algorithm is similar to list recovery introduced in Indyk et al. [2010] and used in Gilbert et al. [2013]. In our algorithm, the messages \mi\ are independent of the hashing, which enables us to obtain a better result.