A Machine Learning Approach to Algorithm Selection for Exact Computation of Treewidth

Borislav Slavchev, Evelina Masliankova, Steven Kelk

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

We present an algorithm selection framework based on machine learning for the exact computation of treewidth, an intensively studied graph parameter that is NP-hard to compute. Specifically, we analyse the comparative performance of three state-of-the-art exact treewidth algorithms on a wide array of graphs and use this information to predict which of the algorithms, on a graph by graph basis, will compute the treewidth the quickest. Experimental results show that the proposed meta-algorithm outperforms existing methods on benchmark instances on all three performance metrics we use: in a nutshell, it computes treewidth faster than any single algorithm in isolation. We analyse our results to derive insights about graph feature importance and the strengths and weaknesses of the algorithms we used. Our results are further evidence of the advantages to be gained by strategically blending machine learning and combinatorial optimisation approaches within a hybrid algorithmic framework. The machine learning model we use is intentionally simple to emphasise that speedup can already be obtained without having to engage in the full complexities of machine learning engineering. We reflect on how future work could extend this simple but effective, proof-of-concept by deploying more sophisticated machine learning models.

Original languageEnglish
Article number200
Number of pages20
JournalAlgorithms
Volume12
Issue number10
DOIs
Publication statusPublished - Sep 2019

Keywords

  • APPROXIMATION
  • COMBINATORIAL OPTIMIZATION
  • PREDICTION
  • algorithm selection
  • combinatorial optimisation
  • machine learning
  • tree decomposition
  • treewidth

Cite this

Slavchev, Borislav ; Masliankova, Evelina ; Kelk, Steven. / A Machine Learning Approach to Algorithm Selection for Exact Computation of Treewidth. In: Algorithms. 2019 ; Vol. 12, No. 10.
@article{24b31abea9e44d87bcf59e08e316ee1e,
title = "A Machine Learning Approach to Algorithm Selection for Exact Computation of Treewidth",
abstract = "We present an algorithm selection framework based on machine learning for the exact computation of treewidth, an intensively studied graph parameter that is NP-hard to compute. Specifically, we analyse the comparative performance of three state-of-the-art exact treewidth algorithms on a wide array of graphs and use this information to predict which of the algorithms, on a graph by graph basis, will compute the treewidth the quickest. Experimental results show that the proposed meta-algorithm outperforms existing methods on benchmark instances on all three performance metrics we use: in a nutshell, it computes treewidth faster than any single algorithm in isolation. We analyse our results to derive insights about graph feature importance and the strengths and weaknesses of the algorithms we used. Our results are further evidence of the advantages to be gained by strategically blending machine learning and combinatorial optimisation approaches within a hybrid algorithmic framework. The machine learning model we use is intentionally simple to emphasise that speedup can already be obtained without having to engage in the full complexities of machine learning engineering. We reflect on how future work could extend this simple but effective, proof-of-concept by deploying more sophisticated machine learning models.",
keywords = "APPROXIMATION, COMBINATORIAL OPTIMIZATION, PREDICTION, algorithm selection, combinatorial optimisation, machine learning, tree decomposition, treewidth",
author = "Borislav Slavchev and Evelina Masliankova and Steven Kelk",
year = "2019",
month = "9",
doi = "10.3390/a12100200",
language = "English",
volume = "12",
journal = "Algorithms",
issn = "1999-4893",
publisher = "MDPI AG",
number = "10",

}

A Machine Learning Approach to Algorithm Selection for Exact Computation of Treewidth. / Slavchev, Borislav; Masliankova, Evelina; Kelk, Steven.

In: Algorithms, Vol. 12, No. 10, 200, 09.2019.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - A Machine Learning Approach to Algorithm Selection for Exact Computation of Treewidth

AU - Slavchev, Borislav

AU - Masliankova, Evelina

AU - Kelk, Steven

PY - 2019/9

Y1 - 2019/9

N2 - We present an algorithm selection framework based on machine learning for the exact computation of treewidth, an intensively studied graph parameter that is NP-hard to compute. Specifically, we analyse the comparative performance of three state-of-the-art exact treewidth algorithms on a wide array of graphs and use this information to predict which of the algorithms, on a graph by graph basis, will compute the treewidth the quickest. Experimental results show that the proposed meta-algorithm outperforms existing methods on benchmark instances on all three performance metrics we use: in a nutshell, it computes treewidth faster than any single algorithm in isolation. We analyse our results to derive insights about graph feature importance and the strengths and weaknesses of the algorithms we used. Our results are further evidence of the advantages to be gained by strategically blending machine learning and combinatorial optimisation approaches within a hybrid algorithmic framework. The machine learning model we use is intentionally simple to emphasise that speedup can already be obtained without having to engage in the full complexities of machine learning engineering. We reflect on how future work could extend this simple but effective, proof-of-concept by deploying more sophisticated machine learning models.

AB - We present an algorithm selection framework based on machine learning for the exact computation of treewidth, an intensively studied graph parameter that is NP-hard to compute. Specifically, we analyse the comparative performance of three state-of-the-art exact treewidth algorithms on a wide array of graphs and use this information to predict which of the algorithms, on a graph by graph basis, will compute the treewidth the quickest. Experimental results show that the proposed meta-algorithm outperforms existing methods on benchmark instances on all three performance metrics we use: in a nutshell, it computes treewidth faster than any single algorithm in isolation. We analyse our results to derive insights about graph feature importance and the strengths and weaknesses of the algorithms we used. Our results are further evidence of the advantages to be gained by strategically blending machine learning and combinatorial optimisation approaches within a hybrid algorithmic framework. The machine learning model we use is intentionally simple to emphasise that speedup can already be obtained without having to engage in the full complexities of machine learning engineering. We reflect on how future work could extend this simple but effective, proof-of-concept by deploying more sophisticated machine learning models.

KW - APPROXIMATION

KW - COMBINATORIAL OPTIMIZATION

KW - PREDICTION

KW - algorithm selection

KW - combinatorial optimisation

KW - machine learning

KW - tree decomposition

KW - treewidth

U2 - 10.3390/a12100200

DO - 10.3390/a12100200

M3 - Article

VL - 12

JO - Algorithms

JF - Algorithms

SN - 1999-4893

IS - 10

M1 - 200

ER -