Presumably Correct Undersampling

Gonzalo Nápoles*, Isel Grau

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Abstract

This paper presents a data pre-processing algorithm to tackle class imbalance in classification problems by undersampling the majority class. It relies on a formalism termed Presumably Correct Decision Sets aimed at isolating easy (presumably correct) and difficult (presumably incorrect) instances in a classification problem. The former are instances with neighbors that largely share their class label, while the latter have neighbors that mostly belong to a different decision class. The proposed algorithm replaces the presumably correct instances belonging to the majority decision class with prototypes, and it operates under the assumption that removing these instances does not change the boundaries of the decision space. Note that this strategy opposes other methods that remove pairs of instances from different classes that are each other’s closest neighbors. We argue that the training and test data should have similar distribution and complexity and that making the decision classes more separable in the training data would only increase the risks of overfitting. The experiments show that our method improves the generalization capabilities of a baseline classifier, while outperforming other undersampling algorithms reported in the literature.

Original languageEnglish
Title of host publicationProgress in Pattern Recognition, Image Analysis, Computer Vision, and Applications
PublisherSpringer Science and Business Media Deutschland GmbH
Pages420-433
Number of pages14
ISBN (Print)9783031490170
DOIs
Publication statusPublished - 2023
Event26th Iberoamerican Congress on Pattern Recognition, CIARP 2023 - Coimbra, Portugal
Duration: 27 Nov 202330 Nov 2023

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14469 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference26th Iberoamerican Congress on Pattern Recognition, CIARP 2023
Country/TerritoryPortugal
CityCoimbra
Period27/11/2330/11/23

Keywords

  • Class Imbalance
  • Pattern Classification
  • Presumably Correct Decision Sets
  • Undersampling

Fingerprint

Dive into the research topics of 'Presumably Correct Undersampling'. Together they form a unique fingerprint.

Cite this