(aside image)

dredviz: dimensionality reduction for information visualization

The dredviz software package implements Neighbor Retrieval Visualizer (NeRV) and Local Multi-Dimensional Scaling (LMDS), dimensionality reduction algorithms recently developed by the Statistical Machine Learning and Bioinformatics group at ICS. Some quality measures for evaluating the quality of a visualization are also included. If you use any of the algorithms or measures, please cite the relevant papers:

NeRV:
Jarkko Venna, Jaakko Peltonen, Kristian Nybo, Helena Aidos, and Samuel Kaski. Information Retrieval Perspective to Nonlinear Dimensionality Reduction for Data Visualization. Journal of Machine Learning Research, 11:451-490, 2010.

Jarkko Venna and Samuel Kaski. Nonlinear Dimensionality Reduction as Information Retrieval. In Marina Meila and Xiaotong Shen, editors, Proceedings of AISTATS 2007, the 11th International Conference on Artificial Intelligence and Statistics. Omnipress, 2007. JMLR Workshop and Conference Proceedings, Volume 2: AISTATS 2007.

LMDS:
Jarkko Venna and Samuel Kaski. Local multidimensional scaling. Neural Networks, 19, pp 889--899, 2006.

We recommend using the Neighbor Retrieval Visualizer NeRV implementation in the package.

The software is written in ANSI/ISO C++ and licensed under the LGPL . Although developed and tested under GNU/Linux, the code should compile and run out of the box on any platform with an ANSI/ISO-compatible C++ compiler.

The current version is 1.0.2, released on 3.10.2012. A few minor bugs were fixed that had no effect on UNIX users but prevented the code from being compiled under Windows. The only notable bug was that the PCA initialization option used the N last principal components instead of the N first ones. In most cases this shouldn't have a significant impact on the final output of the projection algorithm.

Documentation

See the README file included in the package.

Getting and installing the software

Download the tar-ball here, and extract it into an empty directory of your choosing. If you are in a UNIX-like environment and have GNU make and a recent version of g++ installed, simply typing 'make all' at the command line in the directory where you extracted the files should compile everything. See the included README file for instructions on using the software.

Support

This is experimental software provided as is; we welcome any comments and corrections but cannot give any guarantees about the code. If you have any comments or bug reports, please direct them to Kristian Nybo.