Genetic programming with alternative search drivers for detection of retinal blood vessels

Krzysztof Krawiec, Mikołaj Pawlak

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

Abstract

A classification task is a test-based problem, with examples corresponding to tests. A correct classification is equivalent to passing a test, while incorrect to failing it. This applies also to classifying pixels in an image, viz. image segmentation. A natural performance indicator in such a setting is the accuracy of classification, i.e., the fraction of passed tests. When solving a classification tasks with genetic programming, it is thus common to employ this indicator as a fitness function. However, recent developments in GP as well as some earlier work suggest that the quality of evolved solutions may benefit from using other search drivers to guide the traversal of the space of programs. In this study, we systematically verify the usefulness of selected alternative search drivers in the problem of detection of blood vessels in ophthalmology imaging.

Original languageEnglish (US)
Title of host publicationApplications of Evolutionary Computation - 18th European Conference, EvoApplications 2015, Proceedings
EditorsGiovanni Squillero, Antonio M. Mora
PublisherSpringer Verlag
Pages554-566
Number of pages13
ISBN (Electronic)9783319165486
DOIs
StatePublished - 2015
Event18th European Conference on the Applications of Evolutionary Computation, EvoApplications 2015 - Copenhagen, Denmark
Duration: Apr 8 2015Apr 10 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9028
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other18th European Conference on the Applications of Evolutionary Computation, EvoApplications 2015
Country/TerritoryDenmark
CityCopenhagen
Period4/8/154/10/15

Keywords

  • Binary classification
  • Genetic programming
  • Image segmentation
  • Search drivers

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Genetic programming with alternative search drivers for detection of retinal blood vessels'. Together they form a unique fingerprint.

Cite this