Skip to content

Exact MAP-Inference by Confining Combinatorial Search with LP Relaxation

License

Notifications You must be signed in to change notification settings

vislearn/combilp

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

21 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CombiLP

About

This repository contains a re-implementation of the CombiLP algorithm in Python. The original implementation was done by Bogdan Savchynskyy and has been integrated into opengm. I rewrote the whole C++ code base to clean everything up and implemented some major improvements (see references).

You can find more information on the project's website.

References

  1. B. Savchynskyy, J. Kappes, P. Swoboda, C. Schnörr. Global MAP-Optimality by Shrinking the Combinatorial Search Area with Convex Relaxation. NIPS-2013.
  2. S. Haller, P. Swoboda and B. Savchynskyy. Exact MAP-Inference by Confining Combinatorial Search with LP Relaxation. Accepted to AAAI 2018.

Note that this is the not the code used the latest paper, but it should be comparable and much easier to use and hack on, see next section.

Requirements

  • Python 3
  • numpy
  • opengm (optional: to load opengm hdf5 model files)
  • subsolvers you want to use (most of the included as submodule, CPLEX is optional dependency)

About

Exact MAP-Inference by Confining Combinatorial Search with LP Relaxation

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published