Analysis of data-driven parameters in game-theoretic rough sets

  • Authors:

  • Joseph P. Herbert;JingTao Yao

  • Affiliations:

  • Department of Computer Science, University of Regina, Regina, Canada;Department of Computer Science, University of Regina, Regina, Canada

  • Venue:
  • RSKT'11 Proceedings of the 6th international conference on Rough sets and knowledge technology
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

The game-theoretic rough set (GTRS) model provides an alternative approach to the derivation of probabilistic rough set regions. Whereas other models rely on either user-provided parameters or notions of cost for the date set in question, the GTRS model learns these parameters through a game-theoretic process. The parameters can be of the form of probabilities that determine the rough set region bounds or they can be superseded by classification measures whose values represent the current health of the classification system. In this article, we will be analyzing the relationship between the calculated parameters and the learned values of the loss functions. We demonstrate the effectiveness of the game-theoretic rough set model in performing data analysis.