Semi-Supervised Learning through Label Propagation on Geodesics

Please download the codes for Greedy Gradient Max-Cut (GGMC), Gaussian Random Field (GRF),
Local and Global Consistency (LGC) methods at website:
http://www.cs.columbia.edu/~jebara/code.html
Select the "Semi-Supervised Learning Using Greedy Max-Cut CODE"
Uncompress the downloaded file and include it in your path of matlab.
Together with the released codes, one can make preliminary comparisons.
I have to remove dijkstra.mexw64 because it cannot be uploaded to
the matlab exchange system. I replaced dijkstra.mexw64 with dijkstra.cpp
So you can compile it yourself. A really slow implementation using
matlab programming language is also provided, dijkstra.m
However, dijkstra.m is very slow and not recommended.

The codes may take several hours for each demo
Run "Demo_Coil20.m";"Demo_CBCL.m";"Demo_mnist04data.m"
The parameters can be changed.