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

Aram Galstyan; Paul R. Cohen

doi:10.1007/978-3-540-78469-2_13

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

Aram Galstyan, Paul R. Cohen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

5 Scopus citations

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 language	English (US)
Title of host publication	Inductive Logic Programming - 17th International Conference, ILP 2007, Revised Selected Papers
Pages	98-111
Number of pages	14
DOIs	https://doi.org/10.1007/978-3-540-78469-2_13
State	Published - 2008
Externally published	Yes
Event	17th International Conference on Inductive Logic Programming, ILP 2007 - Corvallis, OR, United States Duration: Jun 19 2007 → Jun 21 2007

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	4894 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	17th International Conference on Inductive Logic Programming, ILP 2007
Country/Territory	United States
City	Corvallis, OR
Period	6/19/07 → 6/21/07

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

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

Cite this

Galstyan, A., & Cohen, P. R. (2008). Empirical comparison of "hard" and "soft" label propagation for relational classification. In Inductive Logic Programming - 17th International Conference, ILP 2007, Revised Selected Papers (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.
Inductive Logic Programming - 17th International Conference, ILP 2007, Revised Selected Papers. 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 proceeding › Conference contribution

Galstyan, A & Cohen, PR 2008, Empirical comparison of "hard" and "soft" label propagation for relational classification. in Inductive Logic Programming - 17th International Conference, ILP 2007, Revised Selected Papers. 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 Inductive Logic Programming - 17th International Conference, ILP 2007, Revised Selected Papers. 2008. p. 98-111. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-540-78469-2_13

Galstyan, Aram ; Cohen, Paul R. / Empirical comparison of "hard" and "soft" label propagation for relational classification. Inductive Logic Programming - 17th International Conference, ILP 2007, Revised Selected Papers. 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",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "98--111",

booktitle = "Inductive Logic Programming - 17th International Conference, ILP 2007, Revised Selected Papers",

note = "17th International Conference on Inductive Logic Programming, ILP 2007 ; Conference date: 19-06-2007 Through 21-06-2007",

}

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

AN - SCOPUS:40249091415

SN - 3540784683

SN - 9783540784685

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

SP - 98

EP - 111

BT - Inductive Logic Programming - 17th International Conference, ILP 2007, Revised Selected Papers

T2 - 17th International Conference on Inductive Logic Programming, ILP 2007

Y2 - 19 June 2007 through 21 June 2007

ER -

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

Abstract

Publication series

Other

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this