What is SMOG ?

Smog is a program that induces reduced ordered decision graphs from training set data. It accepts data in one of two popular formats: the ESPRESSO format and the C4.5 format. The current version doesn't accept problems that have continuously valued attributes. Check this page once in a while to see if a version that accepts them comes out. All the files available here are gzipped tar files. Your browser will probably uncompress them, but you will have to untar them.

Binaries

Binary for the alpha architecture
Binary for Intel processors running the linux operating system
Binary for mips boxes running Ultrix
Binary for SUN 4 workstations

Documents

Arlindo L. Oliveira and Alberto Sangiovanni-Vincentelli, Using the Minimum Description Length Principle to Infer Reduced Ordered Decision Graphs, Machine Learning, 25, pp. 23-50, 1996, a journal paper that describes smog in detail.
Arlindo L. Oliveira and Alberto Sangiovanni-Vincentelli, Inferring reduced ordered decision diagrams of minimal description length, Proceedings of the Twelfth International Conference in Machine Learning}, pages 421--429. Morgan Kaufmann, 1995, a shorted version of the previous paper.
The manual for smog.

Data sets

Two small examples in Espresso format.
Two small examples in C4.5 format.
Data sets used for the comparisons made using smog. Get the file, untar it and type make at the top level after changing the first line in the top level Makefile.

Source code

This program is still in an experimental stage and I cannot consider a full release of the source code at this point. The source code is here but is not open to persons outside the CAD group. However, if you are interested in it, please e-mail me to aml@ic.eecs.berkeley.edu or aml@inesc.pt and I will mail it to you..
BDD Libraries that contain the BDD manipulation routines. This is my own version and is known to work. If you have already a copy of the Berkeley BDD package and you trust it, you don't need to get this.