Identifying a non-normal evolving stochastic process based upon the genetic methods

Kangrong Tan, Meifen Chu, Shozo Tokinaga

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In the real world, many evolving stochastic processes appear heavy tails, excess kurtosis, and other non-normal evidences, though, they eventually converge to normals due to the central limit theorem, and the augment effect. So far many studies focusing on the normal cases, such as Brownian Motion, or Geometric Brownian Motion etc, have shown their restrictions in dealing with non-normal phenomena, although they have achieved a great deal of success. Moreover, in many studies, the statistical properties, such as the distributional parameters of an evolving process, have been studied at a special time spot, not having grasped the whole picture during the whole evolving time period. In this paper, we propose to approximate an evolving stochastic process based upon a process characterized by a time-varying mixture distribution family to grasp the whole evolving picture of its evolution behavior. Good statistical properties of such a time-varying process are well illustrated and discussed. The parameters in such a time-varying mixture distribution family are optimized by the Genetic Methods, namely, the Genetic Algorithm (GA) and Genetic Programming (GP). Numerical experiments are carried out and the results prove that our proposed approach works well in dealing with a non-normal evolving stochastic process.

Original languageEnglish
Title of host publicationIntegrated Uncertainty in Knowledge Modelling and Decision Making - International Symposium, IUKM 2011, Proceedings
Pages168-178
Number of pages11
DOIs
Publication statusPublished - Oct 28 2011
Event2011 International Symposium on Integrated Uncertainty in Knowledge Modelling and Decision Making, IUKM 2011 - Hangzhou, China
Duration: Oct 28 2011Oct 30 2011

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7027 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other2011 International Symposium on Integrated Uncertainty in Knowledge Modelling and Decision Making, IUKM 2011
CountryChina
CityHangzhou
Period10/28/1110/30/11

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Identifying a non-normal evolving stochastic process based upon the genetic methods'. Together they form a unique fingerprint.

Cite this