The MATTA tool-set described below consists of tools developed at the University of Houston as well as of tools that were developed by other institutions:

• Classification Tools
  • DELVAUX -- learns naive Bayesian rule-sets for classification tasks relying on an evolutionary programming approach.
  • WOLS-CL -- learns discrimination functions for classification tasks using a symbolic regression / genetic programming approach.
  • C4.5 -- learns decision trees (developed by Quinlan)
  • C4.5_CI -- enhancement of C4.5 that employs constructive induction and other preprocessing techniques to enhance C4.5's prediction accuracy.
  • MASSON -- learns queries that describe commonalities within a given set of objects relying on a genetic programming approach.
  • Tools that employ a "fuzzy engineering" technology
• Clustering Tools
  • ECOBWEB
  • KMEANS
• Dependency Analysis between Variables
  • WOLS-CL -- learns functions using a symbolic regression / genetic programming approach; the use of building block analysis tools is currently explored to enhance the accuracy of the basis system.
  • Regression Analysis Tools (part of the MATLAB statistics tool box)
• Statistical Analysis Tools (we plan to use the MATLAB statistics tool box)
• Fuzzy Engineering Tools (we plan to use the corresponding MATLAB tool box)
• Visualization Tools (we plan to include MATHLAB and possibly some other tools into the MATTA-toolset).
• Knowledge Extraction Tools
  • MAGRITTE (takes a relational database as an input and transform its contents into a "semi-flat" file that can be used as an input of "generalized" classification, clustering, statistical, and visualization algorithms).

last updated: December 11, 1997 at noon