Refining Numerical Constants in First Order Logic Theories

  • Authors:
  • Marco Botta;Roberto Piola

  • Affiliations:
  • Dipartimento di Informatica, Università di Torino, Corso Svizzera 185, 10149 Torino, Italy. botta@di.unito.it;Dipartimento di Informatica, Università di Torino, Corso Svizzera 185, 10149 Torino, Italy. piola@di.unito.it

  • Venue:
  • Machine Learning - Special issue on multistrategy learning
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper proposes a methodfor refining numerical constants occurring in rules of a knowledge baseexpressed in a first order logic language. The method consistsin tuning numerical parameters by performing error gradient descent.The knowledge base to be refined can be manually handcrafted or automaticallyacquired by a symbolic relational learner, able todeal with numerical features. The results of an experimentalanalysis performed on four case studiesshow that the refinement step can be effective in improving classificationperformances.