The weekly SILO Seminar Series is made possible through the generous support of the 3M Company and its Advanced Technology Group


with additional support from the Analytics Group of the Northwestern Mutual Life Insurance Company

Northwestern Mutual

Active Ranking using Pairwise Comparisons | Decomposition Methods for Large Scale LP Decoding

Kevin Jamieson and Siddharth Barman, Graduate Students in ECE and CS respectively

Date and Time: Oct 12, 2011 (12:30 PM)
Location: Orchard room (3280) at the Wisconsin Institute for Discovery Building


Title: Active Ranking using Pairwise Comparisons
by Kevin Jamieson

This talk examines the problem of ranking a collection of objects using pairwise comparisons (rankings of two objects). In general, the ranking of n objects can be identified by standard sorting methods using n log2 n pairwise comparisons. We are interested in natural situations in which relationships among the objects may allow for ranking using far fewer pairwise comparisons. Specifically, we assume that the objects can be embedded into a d-dimensional Euclidean space and that the rankings reflect their relative distances from a common reference point in R^d. We show that under this assumption the number of possible rankings grows like n^{2d} and demonstrate an algorithm that can identify a randomly selected ranking using just slightly more than d log n adaptively selected pairwise comparisons, on average. If instead the comparisons are chosen at random, then almost all pairwise comparisons must be made in order to identify any ranking. In addition, we propose a robust, error-tolerant algorithm that only requires that the pairwise comparisons are probably correct. Experimental studies with synthetic and real datasets support the conclusions of our theoretical analysis.

Title: Decomposition Methods for Large Scale LP Decoding
by Siddharth Barman and Xishuo Liu

Abstract: Feldman et al. showed that linear programming (LP) can be used to decode linear error correcting codes. The bit-error-rate performance of LP decoding is comparable to state-of-the- art decoders based on message passing, but has significantly stronger theoretical guarantees. However, LP decoding when implemented with standard LP solvers does not easily scale to the block lengths of modern error correcting codes. In this talk we draw on decomposition methods from optimization theory to develop efficient distributed algorithms for LP decoding. The key enabling technical result is a nearly linear time algorithm for two-norm projection onto the parity polytope. This allows us to use LP decoding, with all its theoretical guarantees, to decode large-scale error correcting codes efficiently.