Datasets overview

The above gzip package contains 2,757 MAP inference problems, from 29
datasets in computer vision, computational biology, and other fields.
These problems come from 3 original sources [1][2][3].

There are 16 datasets from [1], 10 datasets from [2], and 3 datasets
from [3]. 
[1] originally has 19 datasets, we excluded the 3 datasets:
'correlation-clustering', 'image-seg' and 'seg-3d', 
because they were originally formulated differently (therefore has
very different characteristics) from other MAP inference problems. 

All problems are in the ".map" format. Our MAP inference code package
has description of the format as well as example code for parsing it.


[1] Jörg H. Kappes, Bjoern Andres, Fred A. Hamprecht, Christoph
Schnörr, Sebastian Nowozin, Dhruv Batra, Sungwoong Kim, Bernhard
X. Kausler, Jan Lellmann, Nikos Komodakis, Carsten Rother : "A
Comparative Study of Modern Inference Techniques for Discrete Energy
Minimization Problems", CVPR, 2013
[2] The Probabilistic Inference Challenge (PIC2011)
http://www.cs.huji.ac.il/project/PASCAL/
[3] Chen Yanover, Talya Meltzer, Yair Weiss, Linear Programming
Relaxations and Belief Propagation — An Empirical Study. Journal of
Machine Learning Research, Special Topic on Machine Learning and
Optimization, Sep. 2006