By the same authors

A multi-arm bandit neighbourhood search for routing and scheduling problems

Research output: Contribution to conferencePaper

Standard

A multi-arm bandit neighbourhood search for routing and scheduling problems. / Chen, Yujie; Cowling, Peter Ivan; Polack, Fiona A C; Mourdjis, Philip James.

2016.

Research output: Contribution to conferencePaper

Harvard

Chen, Y, Cowling, PI, Polack, FAC & Mourdjis, PJ 2016, 'A multi-arm bandit neighbourhood search for routing and scheduling problems'.

APA

Chen, Y., Cowling, P. I., Polack, F. A. C., & Mourdjis, P. J. (2016). A multi-arm bandit neighbourhood search for routing and scheduling problems.

Vancouver

Chen Y, Cowling PI, Polack FAC, Mourdjis PJ. A multi-arm bandit neighbourhood search for routing and scheduling problems. 2016.

Author

Chen, Yujie ; Cowling, Peter Ivan ; Polack, Fiona A C ; Mourdjis, Philip James. / A multi-arm bandit neighbourhood search for routing and scheduling problems. 33 p.

Bibtex - Download

@conference{773c01c975b0403cb6a9fd5fa4d9c18a,
title = "A multi-arm bandit neighbourhood search for routing and scheduling problems",
abstract = "Abstract Local search based meta-heuristics such as variable neighbourhoodsearch have achieved remarkable success in solving complex combinatorialproblems. Local search techniques are becoming increasingly popular and areused in a wide variety of meta-heuristics, such as genetic algorithms. Typically,local search iteratively improves a solution by making a series of small moves.Traditionally these methods do not employ any learning mechanism.We treat the selection of a local search neighbourhood as a dynamic multi-armed bandit (D-MAB) problem where learning techniques for solving theD-MAB can be used to guide the local search process. We present a D-MABneighbourhood search (D-MABNS) which can be embedded within any meta-heuristic or hyperheuristic framework. Given a set of neighbourhoods, the aimof D-MABNS is to adapt the search sequence, testing promising solutionsrst. We demonstrate the eectiveness of D-MABNS on two vehicle routingand scheduling problems, the real-world geographically distributed mainte-nance problem (GDMP) and the periodic vehicle routing problem (PVRP).We present comparisons to benchmark instances and give a detailed analysisof parameters, performance and behaviour.Keywords Meta-heuristic Local search Vehicle routing",
author = "Yujie Chen and Cowling, {Peter Ivan} and Polack, {Fiona A C} and Mourdjis, {Philip James}",
year = "2016",
language = "English",

}

RIS (suitable for import to EndNote) - Download

TY - CONF

T1 - A multi-arm bandit neighbourhood search for routing and scheduling problems

AU - Chen, Yujie

AU - Cowling, Peter Ivan

AU - Polack, Fiona A C

AU - Mourdjis, Philip James

PY - 2016

Y1 - 2016

N2 - Abstract Local search based meta-heuristics such as variable neighbourhoodsearch have achieved remarkable success in solving complex combinatorialproblems. Local search techniques are becoming increasingly popular and areused in a wide variety of meta-heuristics, such as genetic algorithms. Typically,local search iteratively improves a solution by making a series of small moves.Traditionally these methods do not employ any learning mechanism.We treat the selection of a local search neighbourhood as a dynamic multi-armed bandit (D-MAB) problem where learning techniques for solving theD-MAB can be used to guide the local search process. We present a D-MABneighbourhood search (D-MABNS) which can be embedded within any meta-heuristic or hyperheuristic framework. Given a set of neighbourhoods, the aimof D-MABNS is to adapt the search sequence, testing promising solutionsrst. We demonstrate the eectiveness of D-MABNS on two vehicle routingand scheduling problems, the real-world geographically distributed mainte-nance problem (GDMP) and the periodic vehicle routing problem (PVRP).We present comparisons to benchmark instances and give a detailed analysisof parameters, performance and behaviour.Keywords Meta-heuristic Local search Vehicle routing

AB - Abstract Local search based meta-heuristics such as variable neighbourhoodsearch have achieved remarkable success in solving complex combinatorialproblems. Local search techniques are becoming increasingly popular and areused in a wide variety of meta-heuristics, such as genetic algorithms. Typically,local search iteratively improves a solution by making a series of small moves.Traditionally these methods do not employ any learning mechanism.We treat the selection of a local search neighbourhood as a dynamic multi-armed bandit (D-MAB) problem where learning techniques for solving theD-MAB can be used to guide the local search process. We present a D-MABneighbourhood search (D-MABNS) which can be embedded within any meta-heuristic or hyperheuristic framework. Given a set of neighbourhoods, the aimof D-MABNS is to adapt the search sequence, testing promising solutionsrst. We demonstrate the eectiveness of D-MABNS on two vehicle routingand scheduling problems, the real-world geographically distributed mainte-nance problem (GDMP) and the periodic vehicle routing problem (PVRP).We present comparisons to benchmark instances and give a detailed analysisof parameters, performance and behaviour.Keywords Meta-heuristic Local search Vehicle routing

M3 - Paper

ER -