Compact MILP models for optimal and Pareto-optimal LAD patterns

Cui Guo, Hong Seo Ryoo

Research output: Contribution to journalArticle

14 Citations (Scopus)

Abstract

This paper develops MILP models for various optimal and Pareto-optimal LAD patterns that involve at most 2n 01 decision variables, where n is the number of support features for the data under analysis, which usually is small. Noting that the previous MILP pattern generation models are defined in 2n+m 01 variables, where m is the number of observations in the dataset with m≫n in general, the new models are expected to generate useful LAD patterns more efficiently. With experiments on six well-studied machine learning datasets, we first demonstrate the efficiency of the new MILP models and next use them to show different utilities of strong prime patterns and strong spanned patterns in enhancing the overall classification accuracy of a LAD decision theory.

Original languageEnglish
Pages (from-to)2339-2348
Number of pages10
JournalDiscrete Applied Mathematics
Volume160
Issue number16-17
DOIs
Publication statusPublished - 2012 Nov 1

Fingerprint

Mixed Integer Linear Programming
Decision theory
Decision Theory
Model
Learning systems
Machine Learning
Experiments
Demonstrate
Experiment

Keywords

  • LAD
  • Maximum prime pattern
  • Maximum spanned pattern
  • MILP
  • Strong prime pattern
  • Strong spanned pattern

ASJC Scopus subject areas

  • Applied Mathematics
  • Discrete Mathematics and Combinatorics

Cite this

Compact MILP models for optimal and Pareto-optimal LAD patterns. / Guo, Cui; Ryoo, Hong Seo.

In: Discrete Applied Mathematics, Vol. 160, No. 16-17, 01.11.2012, p. 2339-2348.

Research output: Contribution to journalArticle

@article{3d83487218c84956bae1c9a41a5f255f,
title = "Compact MILP models for optimal and Pareto-optimal LAD patterns",
abstract = "This paper develops MILP models for various optimal and Pareto-optimal LAD patterns that involve at most 2n 01 decision variables, where n is the number of support features for the data under analysis, which usually is small. Noting that the previous MILP pattern generation models are defined in 2n+m 01 variables, where m is the number of observations in the dataset with m≫n in general, the new models are expected to generate useful LAD patterns more efficiently. With experiments on six well-studied machine learning datasets, we first demonstrate the efficiency of the new MILP models and next use them to show different utilities of strong prime patterns and strong spanned patterns in enhancing the overall classification accuracy of a LAD decision theory.",
keywords = "LAD, Maximum prime pattern, Maximum spanned pattern, MILP, Strong prime pattern, Strong spanned pattern",
author = "Cui Guo and Ryoo, {Hong Seo}",
year = "2012",
month = "11",
day = "1",
doi = "10.1016/j.dam.2012.05.006",
language = "English",
volume = "160",
pages = "2339--2348",
journal = "Discrete Applied Mathematics",
issn = "0166-218X",
publisher = "Elsevier",
number = "16-17",

}

TY - JOUR

T1 - Compact MILP models for optimal and Pareto-optimal LAD patterns

AU - Guo, Cui

AU - Ryoo, Hong Seo

PY - 2012/11/1

Y1 - 2012/11/1

N2 - This paper develops MILP models for various optimal and Pareto-optimal LAD patterns that involve at most 2n 01 decision variables, where n is the number of support features for the data under analysis, which usually is small. Noting that the previous MILP pattern generation models are defined in 2n+m 01 variables, where m is the number of observations in the dataset with m≫n in general, the new models are expected to generate useful LAD patterns more efficiently. With experiments on six well-studied machine learning datasets, we first demonstrate the efficiency of the new MILP models and next use them to show different utilities of strong prime patterns and strong spanned patterns in enhancing the overall classification accuracy of a LAD decision theory.

AB - This paper develops MILP models for various optimal and Pareto-optimal LAD patterns that involve at most 2n 01 decision variables, where n is the number of support features for the data under analysis, which usually is small. Noting that the previous MILP pattern generation models are defined in 2n+m 01 variables, where m is the number of observations in the dataset with m≫n in general, the new models are expected to generate useful LAD patterns more efficiently. With experiments on six well-studied machine learning datasets, we first demonstrate the efficiency of the new MILP models and next use them to show different utilities of strong prime patterns and strong spanned patterns in enhancing the overall classification accuracy of a LAD decision theory.

KW - LAD

KW - Maximum prime pattern

KW - Maximum spanned pattern

KW - MILP

KW - Strong prime pattern

KW - Strong spanned pattern

UR - http://www.scopus.com/inward/record.url?scp=84865084222&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84865084222&partnerID=8YFLogxK

U2 - 10.1016/j.dam.2012.05.006

DO - 10.1016/j.dam.2012.05.006

M3 - Article

AN - SCOPUS:84865084222

VL - 160

SP - 2339

EP - 2348

JO - Discrete Applied Mathematics

JF - Discrete Applied Mathematics

SN - 0166-218X

IS - 16-17

ER -