Matlab Codes for Examples in

Principal Graph and Structure Learning Based on Reversed Graph Embedding

What can this package do?

Given a set of data points, this method can help find a set of new points and the underlying structures over these new points, including curves, bifurcations, self-intersections, loops, and multiple disconnected components.

How to use the code?

The MATLAB codes can be downloaded from online_code, uncompress it and save it on a directory.
NOTE: if you have any problem to download this package, please email to Li Wang via liwang8@uic.edu for request the source code.
In the same directory, type the code in Matlab window as follows.

Synthetic data

>> experiment_toy;

The resulting figures on six synthetic datasets are generated and stored in folder results.

Our method for learning tree structures

Our method for learning L1 graph

Large scale real-world data

>>experiment_large;

The resulting figures on Pendigits data are shown as follows:

Our method for learning tree structures
spanning tree
adjacency matrix

How to cite this paper?

If this package or its datasets are used in your work, please kindly cite the following paper:

Principal Graph and Structure Learning Based on Reversed Graph Embedding. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2016. Accepted.