A simple message passing algorithm for graph partitioning problems

Authors:
Mikael Onsjö;Osamu Watanabe
Affiliations:
Dept. of Computer Sci. and Eng., Chalmers Univ. of Technology, Sweden;Dept. of Math. and Comput. Sci., Tokyo Inst. of Technology, Japan
Venue:
ISAAC'06 Proceedings of the 17th international conference on Algorithms and Computation
Year:
2006

Citing 9
Cited 1

Graph bisection algorithms with good average case behavior

Combinatorica
Probabilistic reasoning in intelligent systems: networks of plausible inference

Probabilistic reasoning in intelligent systems: networks of plausible inference
The solution of some random NP-hard problems in polynomial expected time

Journal of Algorithms
The metropolis algorithm for graph bisection

Discrete Applied Mathematics
Algorithms for graph partitioning on the planted partition model

Random Structures & Algorithms
Permitted and forbidden sets in symmetric threshold-linear networks

Neural Computation
Spectral Partitioning of Random Graphs

FOCS '01 Proceedings of the 42nd IEEE symposium on Foundations of Computer Science
A spectral heuristic for bisecting random graphs

SODA '05 Proceedings of the sixteenth annual ACM-SIAM symposium on Discrete algorithms
Eigenvalues and graph bisection: An average-case analysis

SFCS '87 Proceedings of the 28th Annual Symposium on Foundations of Computer Science

Finding Most Likely Solutions

CiE '07 Proceedings of the 3rd conference on Computability in Europe: Computation and Logic in the Real World

Quantified Score

Hi-index	0.00

Visualization

Abstract

Motivated by the belief propagation, we propose a simple and deterministic message passing algorithm for the Graph Bisection problem and related problems. The running time of the main algorithm is linear w.r.t. the number of vertices and edges. For evaluating its average-case correctness, planted solution models are used. For the Graph Bisection problem under the standard planted solution model with probability parameters p and r, we prove that our algorithm yields a planted solution with probability 1–δ if p–r=Ω(n−1/2log(n/δ)).