29
9

Online Low Rank Matrix Completion

Abstract

We study the problem of {\em online} low-rank matrix completion with M\mathsf{M} users, N\mathsf{N} items and T\mathsf{T} rounds. In each round, the algorithm recommends one item per user, for which it gets a (noisy) reward sampled from a low-rank user-item preference matrix. The goal is to design a method with sub-linear regret (in T\mathsf{T}) and nearly optimal dependence on M\mathsf{M} and N\mathsf{N}. The problem can be easily mapped to the standard multi-armed bandit problem where each item is an {\em independent} arm, but that leads to poor regret as the correlation between arms and users is not exploited. On the other hand, exploiting the low-rank structure of reward matrix is challenging due to non-convexity of the low-rank manifold. We first demonstrate that the low-rank structure can be exploited using a simple explore-then-commit (ETC) approach that ensures a regret of O(polylog(M+N)T2/3)O(\mathsf{polylog} (\mathsf{M}+\mathsf{N}) \mathsf{T}^{2/3}). That is, roughly only polylog(M+N)\mathsf{polylog} (\mathsf{M}+\mathsf{N}) item recommendations are required per user to get a non-trivial solution. We then improve our result for the rank-11 setting which in itself is quite challenging and encapsulates some of the key issues. Here, we propose \textsc{OCTAL} (Online Collaborative filTering using iterAtive user cLustering) that guarantees nearly optimal regret of O(polylog(M+N)T1/2)O(\mathsf{polylog} (\mathsf{M}+\mathsf{N}) \mathsf{T}^{1/2}). OCTAL is based on a novel technique of clustering users that allows iterative elimination of items and leads to a nearly optimal minimax rate.

View on arXiv
Comments on this paper