Journal of KIISE:Software and Applications (한국정보과학회논문지:소프트웨어및응용)

Korean Institute of Information Scientists and Engineers (한국정보과학회)
권호 표지

Analysis of Problem Spaces and Algorithm Behaviors for Feature Selection

특징 선택을 위한 문제 공간과 알고리즘 동작 분석

  • 이진선 (우석대학교 게임콘텐츠학과) ;
  • 오일석 (전북대학교 전자정보공학부)
  • Published : 2006.06.01
Download PDF
⟨ Previous Next ⟩

Abstract

The feature selection algorithms should broadly and efficiently explore the huge problem spaces to find a good solution. This paper attempts to gain insights on the fitness landscape of the spaces and to improve search capability of the algorithms. We investigate the solution spaces in terms of statistics on local maxima and minima. We also analyze behaviors of the existing algorithms and improve their solutions.

특징 선택 알고리즘들은 좋은 해를 찾기 위해 거대한 문제 공간을 폭넓게 효율적으로 탐색하여야 한다. 이 논문에서는 문제 공간의 적합도 지형을 통찰해보고자 하였으며, 알고리즘들의 탐색 능력을 개선하였다. 지역 최고값과 최저값에 대한 통계에 의해 해 공간을 조사한다. 또한 기존 알고리즘들의 동작을 분석하고 이들의 해를 개선하였다.

Keywords

References

  1. J. Kittler, 'Feature selection and extraction,' in Handbook of Pattern Recognition and Image Processing, Academic Press (Edited by T.Y. Young and K.S. Fu), pp.59-83, 1986
  2. I.-S. Oh, J.-S. Lee, and B.-R. Moon, 'Local search-embedded genetic algorithms for feature selection,' Proc. of 16th International Conf. on Pattern Recognition, Vol.II, pp.148-151, 2002 https://doi.org/10.1109/ICPR.2002.1048259
  3. K.D. Boese, A.B. Kahng, and S. Muddu, 'A new adaptive multi-start technique for combinatorial global optimizations,' Operations Research Letters, Vol.16, pp.101-113, 1994 https://doi.org/10.1016/0167-6377(94)90065-5
  4. T. Jones and S. Forrest, 'Fitness distance correlation as a measure of problem difficulty for genetic algorithms,' 6th International Conference on Genetic Algorithms, 1995
  5. Y-H. Kim and B-R. Moon, 'Investigation of the fitness landscapes and multi-parent crossover for graph bipartitioning,' GECCO2003, pp.1123-1135, 2003
  6. P.M. Murphy and D.W. Aha, 'UCI repository for machine learning databases (http://www.ics.uci.edu/~mlearn/MLRepository.html),' Irvine, CA: University of California, Department of Information and Computer Science, 1994
  7. P. Pudil, J. Novovicova, and J. Kittler, 'Floating search methods in feature selection,' Pattern Recognition Letters, Vol.15, pp.1119-1125, 1994 https://doi.org/10.1016/0167-8655(94)90127-9