A kernel method for the optimization of the margin distribution

Fabio Aiolli, Giovanni Martino, Alessandro Sperduti

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

22 Citations (Scopus)

Abstract

Recent results in theoretical machine learning seem to suggest that nice properties of the margin distribution over a training set turns out in a good performance of a classifier. The same principle has been already used in SVM and other kernel based methods as the associated optimization problems try to maximize the minimum of these margins. In this paper, we propose a kernel based method for the direct optimization of the margin distribution (KM-OMD). The method is motivated and analyzed from a game theoretical perspective. A quite efficient optimization algorithm is then proposed. Experimental results over a standard benchmark of 13 datasets have clearly shown state-of-the-art performances.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages305-314
Number of pages10
Volume5163 LNCS
EditionPART 1
DOIs
Publication statusPublished - 2008
Externally publishedYes
Event18th International Conference on Artificial Neural Networks, ICANN 2008 - Prague
Duration: 3 Sep 20086 Sep 2008

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 1
Volume5163 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other18th International Conference on Artificial Neural Networks, ICANN 2008
CityPrague
Period3/9/086/9/08

Fingerprint

Kernel Methods
Margin
Optimization
kernel
Learning systems
Optimization Algorithm
Machine Learning
Classifiers
Efficient Algorithms
Maximise
Classifier
Game
Benchmark
Optimization Problem
Experimental Results

Keywords

  • Game theory
  • Kernel methods
  • Margin distribution

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Aiolli, F., Martino, G., & Sperduti, A. (2008). A kernel method for the optimization of the margin distribution. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 1 ed., Vol. 5163 LNCS, pp. 305-314). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5163 LNCS, No. PART 1). https://doi.org/10.1007/978-3-540-87536-9_32

A kernel method for the optimization of the margin distribution. / Aiolli, Fabio; Martino, Giovanni; Sperduti, Alessandro.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5163 LNCS PART 1. ed. 2008. p. 305-314 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5163 LNCS, No. PART 1).

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

Aiolli, F, Martino, G & Sperduti, A 2008, A kernel method for the optimization of the margin distribution. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 edn, vol. 5163 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 1, vol. 5163 LNCS, pp. 305-314, 18th International Conference on Artificial Neural Networks, ICANN 2008, Prague, 3/9/08. https://doi.org/10.1007/978-3-540-87536-9_32
Aiolli F, Martino G, Sperduti A. A kernel method for the optimization of the margin distribution. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 ed. Vol. 5163 LNCS. 2008. p. 305-314. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1). https://doi.org/10.1007/978-3-540-87536-9_32
Aiolli, Fabio ; Martino, Giovanni ; Sperduti, Alessandro. / A kernel method for the optimization of the margin distribution. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5163 LNCS PART 1. ed. 2008. pp. 305-314 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).
@inproceedings{e99805966ab1481d99d354753e4ff847,
title = "A kernel method for the optimization of the margin distribution",
abstract = "Recent results in theoretical machine learning seem to suggest that nice properties of the margin distribution over a training set turns out in a good performance of a classifier. The same principle has been already used in SVM and other kernel based methods as the associated optimization problems try to maximize the minimum of these margins. In this paper, we propose a kernel based method for the direct optimization of the margin distribution (KM-OMD). The method is motivated and analyzed from a game theoretical perspective. A quite efficient optimization algorithm is then proposed. Experimental results over a standard benchmark of 13 datasets have clearly shown state-of-the-art performances.",
keywords = "Game theory, Kernel methods, Margin distribution",
author = "Fabio Aiolli and Giovanni Martino and Alessandro Sperduti",
year = "2008",
doi = "10.1007/978-3-540-87536-9_32",
language = "English",
isbn = "3540875352",
volume = "5163 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 1",
pages = "305--314",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
edition = "PART 1",

}

TY - GEN

T1 - A kernel method for the optimization of the margin distribution

AU - Aiolli, Fabio

AU - Martino, Giovanni

AU - Sperduti, Alessandro

PY - 2008

Y1 - 2008

N2 - Recent results in theoretical machine learning seem to suggest that nice properties of the margin distribution over a training set turns out in a good performance of a classifier. The same principle has been already used in SVM and other kernel based methods as the associated optimization problems try to maximize the minimum of these margins. In this paper, we propose a kernel based method for the direct optimization of the margin distribution (KM-OMD). The method is motivated and analyzed from a game theoretical perspective. A quite efficient optimization algorithm is then proposed. Experimental results over a standard benchmark of 13 datasets have clearly shown state-of-the-art performances.

AB - Recent results in theoretical machine learning seem to suggest that nice properties of the margin distribution over a training set turns out in a good performance of a classifier. The same principle has been already used in SVM and other kernel based methods as the associated optimization problems try to maximize the minimum of these margins. In this paper, we propose a kernel based method for the direct optimization of the margin distribution (KM-OMD). The method is motivated and analyzed from a game theoretical perspective. A quite efficient optimization algorithm is then proposed. Experimental results over a standard benchmark of 13 datasets have clearly shown state-of-the-art performances.

KW - Game theory

KW - Kernel methods

KW - Margin distribution

UR - http://www.scopus.com/inward/record.url?scp=58849136358&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=58849136358&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-87536-9_32

DO - 10.1007/978-3-540-87536-9_32

M3 - Conference contribution

SN - 3540875352

SN - 9783540875352

VL - 5163 LNCS

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

SP - 305

EP - 314

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

ER -