Empirical comparison of "hard" and "soft" label propagation for relational classification

Aram Galstyan, Paul R Cohen

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

1 Citation (Scopus)

Abstract

In this paper we differentiate between hard and soft label propagation for classification of relational (networked) data. The latter method assigns probabilities or class-membership scores to data instances, then propagates these scores throughout the networked data, whereas the former works by explicitly propagating class labels at each iteration. We present a comparative empirical study of these methods applied to a relational binary classification task, and evaluate two approaches on both synthetic and real-world relational data. Our results indicate that while neither approach dominates the other over the entire range of input data parameters, there are some interesting and non-trivial tradeoffs between them.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages98-111
Number of pages14
Volume4894 LNAI
DOIs
StatePublished - 2008
Externally publishedYes
Event17th International Conference on Inductive Logic Programming, ILP 2007 - Corvallis, OR, United States
Duration: Jun 19 2007Jun 21 2007

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4894 LNAI
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other17th International Conference on Inductive Logic Programming, ILP 2007
CountryUnited States
CityCorvallis, OR
Period6/19/076/21/07

Fingerprint

Labels
Propagation
Binary Classification
Differentiate
Empirical Study
Assign
Trade-offs
Entire
Iteration
Evaluate
Range of data
Class

ASJC Scopus subject areas

  • Computer Science(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Theoretical Computer Science

Cite this

Galstyan, A., & Cohen, P. R. (2008). Empirical comparison of "hard" and "soft" label propagation for relational classification. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 4894 LNAI, pp. 98-111). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4894 LNAI). https://doi.org/10.1007/978-3-540-78469-2_13

Empirical comparison of "hard" and "soft" label propagation for relational classification. / Galstyan, Aram; Cohen, Paul R.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4894 LNAI 2008. p. 98-111 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4894 LNAI).

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

Galstyan, A & Cohen, PR 2008, Empirical comparison of "hard" and "soft" label propagation for relational classification. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 4894 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 4894 LNAI, pp. 98-111, 17th International Conference on Inductive Logic Programming, ILP 2007, Corvallis, OR, United States, 6/19/07. https://doi.org/10.1007/978-3-540-78469-2_13
Galstyan A, Cohen PR. Empirical comparison of "hard" and "soft" label propagation for relational classification. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4894 LNAI. 2008. p. 98-111. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-540-78469-2_13
Galstyan, Aram ; Cohen, Paul R. / Empirical comparison of "hard" and "soft" label propagation for relational classification. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4894 LNAI 2008. pp. 98-111 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{c17ef8925ff744dcae7af09713a9046a,
title = "Empirical comparison of {"}hard{"} and {"}soft{"} label propagation for relational classification",
abstract = "In this paper we differentiate between hard and soft label propagation for classification of relational (networked) data. The latter method assigns probabilities or class-membership scores to data instances, then propagates these scores throughout the networked data, whereas the former works by explicitly propagating class labels at each iteration. We present a comparative empirical study of these methods applied to a relational binary classification task, and evaluate two approaches on both synthetic and real-world relational data. Our results indicate that while neither approach dominates the other over the entire range of input data parameters, there are some interesting and non-trivial tradeoffs between them.",
author = "Aram Galstyan and Cohen, {Paul R}",
year = "2008",
doi = "10.1007/978-3-540-78469-2_13",
language = "English (US)",
isbn = "3540784683",
volume = "4894 LNAI",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "98--111",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Empirical comparison of "hard" and "soft" label propagation for relational classification

AU - Galstyan, Aram

AU - Cohen, Paul R

PY - 2008

Y1 - 2008

N2 - In this paper we differentiate between hard and soft label propagation for classification of relational (networked) data. The latter method assigns probabilities or class-membership scores to data instances, then propagates these scores throughout the networked data, whereas the former works by explicitly propagating class labels at each iteration. We present a comparative empirical study of these methods applied to a relational binary classification task, and evaluate two approaches on both synthetic and real-world relational data. Our results indicate that while neither approach dominates the other over the entire range of input data parameters, there are some interesting and non-trivial tradeoffs between them.

AB - In this paper we differentiate between hard and soft label propagation for classification of relational (networked) data. The latter method assigns probabilities or class-membership scores to data instances, then propagates these scores throughout the networked data, whereas the former works by explicitly propagating class labels at each iteration. We present a comparative empirical study of these methods applied to a relational binary classification task, and evaluate two approaches on both synthetic and real-world relational data. Our results indicate that while neither approach dominates the other over the entire range of input data parameters, there are some interesting and non-trivial tradeoffs between them.

UR - http://www.scopus.com/inward/record.url?scp=40249091415&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=40249091415&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-78469-2_13

DO - 10.1007/978-3-540-78469-2_13

M3 - Conference contribution

SN - 3540784683

SN - 9783540784685

VL - 4894 LNAI

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 98

EP - 111

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -