To main content

Efficient mining of pareto-front high expected utility patterns

Abstract

In this paper, we present a model called MHEUPM to efficiently mine the interesting high expected utility patterns (HEUPs) by employing the multi-objective evolutionary framework. The model considers both uncertainty and utility factors to discover meaningful HEUPMs without requiring pre-defined threshold values (such as minimum utility and minimum uncertainty). The effectiveness of the model is validated using two encoding methodologies. The proposed MHEUPM model can discover a set of HEUPs within a limited period. The efficiency of the proposed model is determined through rigorous analysis and compared to the standard pattern-mining methods in terms of hypervolume, convergence, and number of the discovered patterns.

Category

Academic article

Language

English

Author(s)

  • Usman Ahmed
  • Jerry Chun-Wei Lin
  • Jimmy Ming-Tai Wu
  • Youcef Djenouri
  • Gautam Srivastava
  • Suresh Kumar Mukhiya

Affiliation

  • SINTEF Digital / Mathematics and Cybernetics
  • Western Norway University of Applied Sciences
  • Shandong University of Technology
  • Brandon University

Year

2020

Published in

Lecture Notes in Computer Science (LNCS)

ISSN

0302-9743

Volume

12144

Page(s)

872 - 883

View this publication at Norwegian Research Information Repository