Red Bayesiana


EDAs applied to the Table Ordering Problem


Last revision, November 2008

©2008 ISG group (UPV/EHU) & UPM


Researchers from the Intelligent Systems Group of the University of the Basque Country (UPV/EHU) and the Technical University of Madrid (UPM) have developed several theoretical concepts of Estimation of Distribution Algorithms (EDAs), as well as a program which implements EDAs and some more appeared in the literature.

This concrete web page addresses the application of EDAs to the Table Ordering Problem: given the information of a 2D table, it optimizes the arrangement of rows and column to return a table easier to interpret. The authors of the study are Endika Bengoetxea from UPV/EHU, and Pedro Larrañaga, Concha Bielza and Juan Antonio Sánchez Pozo from the UPM.

The next is a program which applies by default the best parameters for this problem accoording to our research:

Windows NT/2000/XP version: CEDA-OrderingOfTables.exe

To run it, from the Start->execute option, type cmd; from it, type:

CEDA-OrderingOfTables

and you will have on the screen the whole list of parameters and options accepted by the program. For instance, a typical execution, with UMDA, would be started in the following way:

CEDA-OrderingOfTables -l0 -o out.dat

which would leave the results in the file out.dat, although these will also appear on the screen.

Trying it with a concrete table

The tables need to have a concrete format to be accepted by our program. cities-4.tabla is an example file that illustrates the format required.

If we execute CEDA-OrderingOfTables -F cities-4.tabla -o out.dat we would obtain This outputs the process on screen (and on the file Screen.txt).

The program also returns a new table file with the same format with the name out-EDA.tabla.

The ISG reserach group also distributes the source code of EDA free of charge to the scientific community. For more ifnormation about it press go to EDA source code web page

Web page of the ISG group | Email us about comments/more information/bugs


Web page maintained by Endika Bengoetxea <endika@ehu.es>