Skip to main navigation Skip to search Skip to main content

Dynamic selection approaches for multiple classifier systems

  • Universidade Federal do Tocantins
  • Concordia University

Research output: Contribution to journal › Journal Article › peer-review

101 Citations (Scopus)

Abstract

In this paper we propose a new approach for dynamic selection of ensembles of classifiers. Based on the concept named multistage organizations, the main objective of which is to define a multi-layer fusion function adapted to each recognition problem, we propose dynamic multistage organization (DMO), which defines the best multistage structure for each test sample. By extending Dos Santos et al.'s approach, we propose two implementations for DMO, namely DSAm and DSAc. While the former considers a set of dynamic selection functions to generalize a DMO structure, the latter considers contextual information, represented by the output profiles computed from the validation dataset, to conduct this task. The experimental evaluation, considering both small and large datasets, demonstrated that DSAc dominated DSAm on most problems, showing that the use of contextual information can reach better performance than other existing methods. In addition, the performance of DSAc can also be enhanced in incremental learning. However, the most important observation, supported by additional experiments, is that dynamic selection is generally preferred over static approaches when the recognition problem presents a high level of uncertainty.

Original languageEnglish
Pages (from-to)673-688
Number of pages16
JournalNeural Computing and Applications
Volume22
Issue number3-4
DOIs
Publication statusPublished - Mar 2013

Keywords

  • Adaptive system
  • Dynamic selection
  • Ensembles of classifiers
  • Incremental learning
  • Multiple classifier systems
  • Multistage organizations

ASJC Scopus subject areas

  • Software
  • Artificial Intelligence

Fingerprint

Dive into the research topics of 'Dynamic selection approaches for multiple classifier systems'. These topics are generated from the title and abstract of the publication. Together, they form a unique fingerprint.

Cite this