1: \begin{abstract}

2: Symmetric Nonnegative Matrix Factorization (SNMF) models arise naturally

3: as simple reformulations of many standard clustering algorithms including

4: the popular spectral clustering method. Recent work has demonstrated

5: that an elementary instance of SNMF provides superior clustering quality

6: compared to many classic clustering algorithms on a variety of synthetic

7: and real world data sets. In this work, we present novel reformulations

8: of this instance of SNMF based on the notion of variable splitting

9: and produce two fast and effective algorithms for its optimization

10: using i) the provably convergent Accelerated Proximal Gradient (APG)

11: procedure and ii) a heuristic version of the Alternating Direction

12: Method of Multipliers (ADMM) framework.  Our two algorithms present an interesting tradeoff between computational speed and mathematical convergence guarantee: while the former method is provably convergent it is considerably slower than the latter approach, for which we also provide significant but less stringent mathematical proof regarding its convergence.  Through extensive experiments

13: we show not only that the efficacy of these approaches is equal to

14: that of the state of the art SNMF algorithm, but also that the latter

15: of our algorithms is extremely fast being one to two orders of magnitude

16: faster in terms of total computation time than the state of the art

17: approach, outperforming even spectral clustering in terms of computation

18: time on large data sets.

19: \end{abstract}