Fuzzy Knowledge Construction from Statistical Data

The object of this research activity is to investigate methods and techniques to construct membership functions for each linguistic variable involved in the rules which a fuzzy DSS deals with, starting from statistical data. Our approach is meant to investigate the links potentially existing between fuzzy logic theory, possibility theory, probability theory and the observation of the frequency of events.

At the theoretical level, we are involved in the investigation of a set of transformations of a probability function into a fuzzy subset interpreted as a possibility function, with the final aim of losing as little information as possible. At the same time, we are also exploring methods for directly determining fuzzy interpretations of statistical/probabilistic data sets by defining reference limits of membership functions through the application of fractiles, confidence intervals, and other well-known concepts and results in statistics.

Publications

  • M. Pota, M. Esposito, G. De Pietro, Transformation of probability distribution into fuzzy set interpretable with likelihood view. In 11th International Conference On Hybrid Intelligent Systems, December 5-8, Malacca, Malaysia.
  • M. Pota, M. Esposito, G. De Pietro, Properties Evaluation of an Approach Based on Probability-Possibility Transformation. In CISSE11: Proceedings of the International Joint Conferences on Computer, Information, and Systems Sciences, and Engineering, December 3-12.
  • M. Pota, M. Esposito and G. De Pietro: From Likelihood Uncertainty to Fuzziness: a Possibility-Based Approach for Building Clinical DSSs The 7th International Conference on Hybrid Artificial Intelligence Systems, Salamanca, Spain, March 2012.