Ordered Decompositional DAG kernels enhancements

Giovanni Martino, Nicolò Navarin, Alessandro Sperduti

Research output: Contribution to journalArticle

12 Citations (Scopus)

Abstract

In this paper, we show how the Ordered Decomposition DAGs (ODD) kernel framework, a framework that allows the definition of graph kernels from tree kernels, allows to easily define new state-of-the-art graph kernels. Here we consider a fast graph kernel based on the Subtree kernel (ST), and we propose various enhancements to increase its expressiveness. The proposed DAG kernel has the same worst-case complexity as the one based on ST, but an improved expressivity due to an augmented set of features. Moreover, we propose a novel weighting scheme for the features, which can be applied to other kernels of the ODD framework. These improvements allow the proposed kernels to improve on the classification performances of the ST-based kernel for several real-world datasets, reaching state-of-the-art performances.

Original languageEnglish
JournalNeurocomputing
DOIs
Publication statusAccepted/In press - 12 Jul 2015

Fingerprint

Decomposition
Datasets

Keywords

  • Classification
  • Graph kernels
  • Kernel functions
  • Kernel methods

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Cognitive Neuroscience

Cite this

Ordered Decompositional DAG kernels enhancements. / Martino, Giovanni; Navarin, Nicolò; Sperduti, Alessandro.

In: Neurocomputing, 12.07.2015.

Research output: Contribution to journalArticle

Martino, Giovanni ; Navarin, Nicolò ; Sperduti, Alessandro. / Ordered Decompositional DAG kernels enhancements. In: Neurocomputing. 2015.
@article{f2abc8ccb90b4bdebcac83d7e955a476,
title = "Ordered Decompositional DAG kernels enhancements",
abstract = "In this paper, we show how the Ordered Decomposition DAGs (ODD) kernel framework, a framework that allows the definition of graph kernels from tree kernels, allows to easily define new state-of-the-art graph kernels. Here we consider a fast graph kernel based on the Subtree kernel (ST), and we propose various enhancements to increase its expressiveness. The proposed DAG kernel has the same worst-case complexity as the one based on ST, but an improved expressivity due to an augmented set of features. Moreover, we propose a novel weighting scheme for the features, which can be applied to other kernels of the ODD framework. These improvements allow the proposed kernels to improve on the classification performances of the ST-based kernel for several real-world datasets, reaching state-of-the-art performances.",
keywords = "Classification, Graph kernels, Kernel functions, Kernel methods",
author = "Giovanni Martino and Nicol{\`o} Navarin and Alessandro Sperduti",
year = "2015",
month = "7",
day = "12",
doi = "10.1016/j.neucom.2015.12.110",
language = "English",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",

}

TY - JOUR

T1 - Ordered Decompositional DAG kernels enhancements

AU - Martino, Giovanni

AU - Navarin, Nicolò

AU - Sperduti, Alessandro

PY - 2015/7/12

Y1 - 2015/7/12

N2 - In this paper, we show how the Ordered Decomposition DAGs (ODD) kernel framework, a framework that allows the definition of graph kernels from tree kernels, allows to easily define new state-of-the-art graph kernels. Here we consider a fast graph kernel based on the Subtree kernel (ST), and we propose various enhancements to increase its expressiveness. The proposed DAG kernel has the same worst-case complexity as the one based on ST, but an improved expressivity due to an augmented set of features. Moreover, we propose a novel weighting scheme for the features, which can be applied to other kernels of the ODD framework. These improvements allow the proposed kernels to improve on the classification performances of the ST-based kernel for several real-world datasets, reaching state-of-the-art performances.

AB - In this paper, we show how the Ordered Decomposition DAGs (ODD) kernel framework, a framework that allows the definition of graph kernels from tree kernels, allows to easily define new state-of-the-art graph kernels. Here we consider a fast graph kernel based on the Subtree kernel (ST), and we propose various enhancements to increase its expressiveness. The proposed DAG kernel has the same worst-case complexity as the one based on ST, but an improved expressivity due to an augmented set of features. Moreover, we propose a novel weighting scheme for the features, which can be applied to other kernels of the ODD framework. These improvements allow the proposed kernels to improve on the classification performances of the ST-based kernel for several real-world datasets, reaching state-of-the-art performances.

KW - Classification

KW - Graph kernels

KW - Kernel functions

KW - Kernel methods

UR - http://www.scopus.com/inward/record.url?scp=84960539523&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84960539523&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2015.12.110

DO - 10.1016/j.neucom.2015.12.110

M3 - Article

AN - SCOPUS:84960539523

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

ER -