Parallel Multicut Segmentation via Dual Decomposition

  title={Parallel Multicut Segmentation via Dual Decomposition},
  author={Julian Yarkony and Thorsten Beier and Pierre Baldi and Fred A. Hamprecht},
We propose a new outer relaxation of the multicut polytope, along with a dual decomposition approach for correlation clustering and multicut segmentation, for general graphs. Each subproblem is a minimum st-cut problem and can thus be solved efficiently. An optimal reparameterization is found using subgradients and affords a new characterization of the basic LP relaxation of the multicut problem, as well as informed decoding heuristics. The algorithm we propose for solving the problem… 

Massively Parallel Benders Decomposition for Correlation Clustering

The Benders decomposition approach provides a promising new avenue to accelerate optimization for CC, and allows for massive parallelization.

A Benders Decomposition Approach to Correlation Clustering

  • Jovita LukasikMargret KeuperM. SinghJulian Yarkony
  • Computer Science
    2020 IEEE/ACM Workshop on Machine Learning in High Performance Computing Environments (MLHPC) and Workshop on Artificial Intelligence and Machine Learning for Scientific Applications (AI4S)
  • 2020
The Benders decomposition approach provides a promising new avenue to accelerate optimization for CC, and, in contrast to previous cutting plane approaches, theoretically allows for massive parallelization.

Planar Ultrametric Rounding for Image Segmentation

A dual cutting plane scheme is introduced that uses minimum cost perfect matching as a subroutine in order to efficiently explore the space of planar partitions and applies this algorithm to the problem of hierarchical image segmentation.

Planar Ultrametrics for Image Segmentation

We study the problem of hierarchical clustering on planar graphs. We formulate this in terms of finding the closest ultrametric to a specified set of distances and solve it using an LP relaxation

A Message Passing Algorithm for the Minimum Cost Multicut Problem

  • P. SwobodaBjoern Andres
  • Computer Science
    2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2017
An algorithm that alternates between message passing and efficient separation of cycle-and odd-wheel inequalities is defined, which is more efficient than state-of-the-art algorithms based on linear programming.

Partial Optimality and Fast Lower Bounds for Weighted Correlation Clustering

A re-weighting with the dual solution by which efficient local search algorithms converge to better feasible solutions to solve the problem for series-parallel graphs to optimality, in linear time.

Flow-Partitionable Signed Graphs

The class of flow-partitionable signed graphs is defined, which have the property that the standard linear programming relaxation based on so-called cycle inequalities is tight, and satisfy an exact max-multiflow-min-multicut relation in the associated instances of minimum multicut.



Segmenting Planar Superpixel Adjacency Graphs w.r.t. Non-planar Superpixel Affinity Graphs

This work considers minimum multicuts of superpixel affinity graphs in which all affinities between non-adjacent superpixels are negative and proposes a relaxation by Lagrangian decomposition and a constrained set of re-parameterizations for which it can optimize exactly and efficiently.

Fast Planar Correlation Clustering for Image Segmentation

This work describes a new optimization scheme for finding high-quality clusterings in planar graphs that uses weighted perfect matching as a subroutine and demonstrates this approach on the problem of image segmentation where this approach outperforms existing global optimization techniques in minimizing the objective.

Probabilistic image segmentation with closedness constraints

We propose a novel graphical model for probabilistic image segmentation that contributes both to aspects of perceptual grouping in connection with image segmentation, and to globally optimal

1 Introduction to Dual Decomposition for Inference

This chapter introduces the general technique of dual decomposition through its application to the problem of finding the most likely (MAP) assignment in Markov random fields, and discusses both subgradient and block coordinate descent approaches to solving the dual problem.

Globally Optimal Closed-Surface Segmentation for Connectomics

A practical cutting-plane approach to solve the MAP inference problem to global optimality despite its NP-hardness is proposed and this approach is applied to challenging large-scale 3D segmentation problems for neural circuit reconstruction (Connectomics), demonstrating the advantage of this higher-order model over independent decisions and finite-order approximations.

New Outer Bounds on the Marginal Polytope

A new class of outer bounds on the marginal polytope is given, and a cutting-plane algorithm for efficiently optimizing over these constraints is proposed, which gives a new variational inference algorithm for probabilistic inference in discrete Markov Random Fields (MRFs).

MRF Energy Minimization and Beyond via Dual Decomposition

It is shown that by appropriately choosing what subproblems to use, one can design novel and very powerful MRF optimization algorithms, which are able to derive algorithms that generalize and extend state-of-the-art message-passing methods, and take full advantage of the special structure that may exist in particular MRFs.

The Complexity of Multiterminal Cuts

It is shown that the problem becomes NP-hard as soon as $k=3$, but can be solved in polynomial time for planar graphs for any fixed $k$, if the planar problem is NP- hard, however, if £k$ is not fixed.

Optimizing Binary MRFs via Extended Roof Duality

An efficient implementation of the "probing" technique is discussed, which simplifies the MRF while preserving the global optimum, and a new technique which takes an arbitrary input labeling and tries to improve its energy is presented.

Dynamic Graph Cuts for Efficient Inference in Markov Random Fields

A fast new fully dynamic algorithm for the st-mincut/max-flow problem and how this algorithm can be used to efficiently compute MAP solutions for certain dynamically changing MRF models in computer vision such as image segmentation is shown.