Fast support vector machines for structural kernels

Aliaksei Severyn, Alessandro Moschitti

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

7 Citations (Scopus)

Abstract

In this paper, we propose three important enhancements of the approximate cutting plane algorithm (CPA) to train Support Vector Machines with structural kernels: (i) we exploit a compact yet exact representation of cutting plane models using directed acyclic graphs to speed up both training and classification, (ii) we provide a parallel implementation, which makes the training scale almost linearly with the number of CPUs, and (iii) we propose an alternative sampling strategy to handle class-imbalanced problem and show that theoretical convergence bounds are preserved. The experimental evaluations on three diverse datasets demonstrate the soundness of our approach and the possibility to carry out fast learning and classification with structural kernels.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages175-190
Number of pages16
Volume6913 LNAI
EditionPART 3
DOIs
Publication statusPublished - 9 Sep 2011
Externally publishedYes
EventEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2011 - Athens, Greece
Duration: 5 Sep 20119 Sep 2011

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 3
Volume6913 LNAI
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

OtherEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2011
CountryGreece
CityAthens
Period5/9/119/9/11

Fingerprint

Support vector machines
Support Vector Machine
kernel
Cutting Plane Algorithm
Sampling Strategy
Cutting Planes
Approximate Algorithm
Directed Acyclic Graph
Soundness
Parallel Implementation
Experimental Evaluation
Program processors
Speedup
Enhancement
Linearly
Sampling
Alternatives
Demonstrate
Training
Model

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Severyn, A., & Moschitti, A. (2011). Fast support vector machines for structural kernels. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 3 ed., Vol. 6913 LNAI, pp. 175-190). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6913 LNAI, No. PART 3). https://doi.org/10.1007/978-3-642-23808-6_12

Fast support vector machines for structural kernels. / Severyn, Aliaksei; Moschitti, Alessandro.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 6913 LNAI PART 3. ed. 2011. p. 175-190 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6913 LNAI, No. PART 3).

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

Severyn, A & Moschitti, A 2011, Fast support vector machines for structural kernels. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 3 edn, vol. 6913 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 3, vol. 6913 LNAI, pp. 175-190, European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2011, Athens, Greece, 5/9/11. https://doi.org/10.1007/978-3-642-23808-6_12
Severyn A, Moschitti A. Fast support vector machines for structural kernels. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 3 ed. Vol. 6913 LNAI. 2011. p. 175-190. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 3). https://doi.org/10.1007/978-3-642-23808-6_12
Severyn, Aliaksei ; Moschitti, Alessandro. / Fast support vector machines for structural kernels. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 6913 LNAI PART 3. ed. 2011. pp. 175-190 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 3).
@inproceedings{8e593c7e4b5949faa584319f306626c4,
title = "Fast support vector machines for structural kernels",
abstract = "In this paper, we propose three important enhancements of the approximate cutting plane algorithm (CPA) to train Support Vector Machines with structural kernels: (i) we exploit a compact yet exact representation of cutting plane models using directed acyclic graphs to speed up both training and classification, (ii) we provide a parallel implementation, which makes the training scale almost linearly with the number of CPUs, and (iii) we propose an alternative sampling strategy to handle class-imbalanced problem and show that theoretical convergence bounds are preserved. The experimental evaluations on three diverse datasets demonstrate the soundness of our approach and the possibility to carry out fast learning and classification with structural kernels.",
author = "Aliaksei Severyn and Alessandro Moschitti",
year = "2011",
month = "9",
day = "9",
doi = "10.1007/978-3-642-23808-6_12",
language = "English",
isbn = "9783642238079",
volume = "6913 LNAI",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 3",
pages = "175--190",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
edition = "PART 3",

}

TY - GEN

T1 - Fast support vector machines for structural kernels

AU - Severyn, Aliaksei

AU - Moschitti, Alessandro

PY - 2011/9/9

Y1 - 2011/9/9

N2 - In this paper, we propose three important enhancements of the approximate cutting plane algorithm (CPA) to train Support Vector Machines with structural kernels: (i) we exploit a compact yet exact representation of cutting plane models using directed acyclic graphs to speed up both training and classification, (ii) we provide a parallel implementation, which makes the training scale almost linearly with the number of CPUs, and (iii) we propose an alternative sampling strategy to handle class-imbalanced problem and show that theoretical convergence bounds are preserved. The experimental evaluations on three diverse datasets demonstrate the soundness of our approach and the possibility to carry out fast learning and classification with structural kernels.

AB - In this paper, we propose three important enhancements of the approximate cutting plane algorithm (CPA) to train Support Vector Machines with structural kernels: (i) we exploit a compact yet exact representation of cutting plane models using directed acyclic graphs to speed up both training and classification, (ii) we provide a parallel implementation, which makes the training scale almost linearly with the number of CPUs, and (iii) we propose an alternative sampling strategy to handle class-imbalanced problem and show that theoretical convergence bounds are preserved. The experimental evaluations on three diverse datasets demonstrate the soundness of our approach and the possibility to carry out fast learning and classification with structural kernels.

UR - http://www.scopus.com/inward/record.url?scp=80052419154&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80052419154&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-23808-6_12

DO - 10.1007/978-3-642-23808-6_12

M3 - Conference contribution

AN - SCOPUS:80052419154

SN - 9783642238079

VL - 6913 LNAI

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

SP - 175

EP - 190

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

ER -