Learning large neighborhood search policy for integer programming

We propose a deep reinforcement learning (RL) method to learn large neighborhood search (LNS) policy for integer programming (IP). The RL policy is trained as the destroy operator to select a subset of variables at each step, which is reoptimized by an IP solver as the repair operator. However, the...

Full description

Saved in:
Bibliographic Details
Main Authors: WU, Yaoxin, SONG, Wen, CAO, Zhiguang, ZHANG, Jie
Format: text
Language:English
Published: Institutional Knowledge at Singapore Management University 2021
Subjects:
Online Access:https://ink.library.smu.edu.sg/sis_research/8159
https://ink.library.smu.edu.sg/context/sis_research/article/9162/viewcontent/NeurIPS_2021_learning_large_neighborhood_search_policy_for_integer_programming_Paper.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Singapore Management University
Language: English
id sg-smu-ink.sis_research-9162
record_format dspace
spelling sg-smu-ink.sis_research-91622023-09-26T10:38:39Z Learning large neighborhood search policy for integer programming WU, Yaoxin SONG, Wen CAO, Zhiguang ZHANG, Jie We propose a deep reinforcement learning (RL) method to learn large neighborhood search (LNS) policy for integer programming (IP). The RL policy is trained as the destroy operator to select a subset of variables at each step, which is reoptimized by an IP solver as the repair operator. However, the combinatorial number of variable subsets prevents direct application of typical RL algorithms. To tackle this challenge, we represent all subsets by factorizing them into binary decisions on each variable. We then design a neural network to learn policies for each variable in parallel, trained by a customized actor-critic algorithm. We evaluate the proposed method on four representative IP problems. Results show that it can find better solutions than SCIP in much less time, and significantly outperform other LNS baselines with the same runtime. Moreover, these advantages notably persist when the policies generalize to larger problems. Further experiments with Gurobi also reveal that our method can outperform this state-of-the-art commercial solver within the same time limit. 2021-12-01T08:00:00Z text application/pdf https://ink.library.smu.edu.sg/sis_research/8159 info:doi/10.48550/arXiv.2111.03466 https://ink.library.smu.edu.sg/context/sis_research/article/9162/viewcontent/NeurIPS_2021_learning_large_neighborhood_search_policy_for_integer_programming_Paper.pdf http://creativecommons.org/licenses/by-nc-nd/4.0/ Research Collection School Of Computing and Information Systems eng Institutional Knowledge at Singapore Management University Actor-critic algorithm Binary decision Integer programming problems Neural-networks Reinforcement learning algorithms Databases and Information Systems
institution Singapore Management University
building SMU Libraries
continent Asia
country Singapore
Singapore
content_provider SMU Libraries
collection InK@SMU
language English
topic Actor-critic algorithm
Binary decision
Integer programming problems
Neural-networks
Reinforcement learning algorithms
Databases and Information Systems
spellingShingle Actor-critic algorithm
Binary decision
Integer programming problems
Neural-networks
Reinforcement learning algorithms
Databases and Information Systems
WU, Yaoxin
SONG, Wen
CAO, Zhiguang
ZHANG, Jie
Learning large neighborhood search policy for integer programming
description We propose a deep reinforcement learning (RL) method to learn large neighborhood search (LNS) policy for integer programming (IP). The RL policy is trained as the destroy operator to select a subset of variables at each step, which is reoptimized by an IP solver as the repair operator. However, the combinatorial number of variable subsets prevents direct application of typical RL algorithms. To tackle this challenge, we represent all subsets by factorizing them into binary decisions on each variable. We then design a neural network to learn policies for each variable in parallel, trained by a customized actor-critic algorithm. We evaluate the proposed method on four representative IP problems. Results show that it can find better solutions than SCIP in much less time, and significantly outperform other LNS baselines with the same runtime. Moreover, these advantages notably persist when the policies generalize to larger problems. Further experiments with Gurobi also reveal that our method can outperform this state-of-the-art commercial solver within the same time limit.
format text
author WU, Yaoxin
SONG, Wen
CAO, Zhiguang
ZHANG, Jie
author_facet WU, Yaoxin
SONG, Wen
CAO, Zhiguang
ZHANG, Jie
author_sort WU, Yaoxin
title Learning large neighborhood search policy for integer programming
title_short Learning large neighborhood search policy for integer programming
title_full Learning large neighborhood search policy for integer programming
title_fullStr Learning large neighborhood search policy for integer programming
title_full_unstemmed Learning large neighborhood search policy for integer programming
title_sort learning large neighborhood search policy for integer programming
publisher Institutional Knowledge at Singapore Management University
publishDate 2021
url https://ink.library.smu.edu.sg/sis_research/8159
https://ink.library.smu.edu.sg/context/sis_research/article/9162/viewcontent/NeurIPS_2021_learning_large_neighborhood_search_policy_for_integer_programming_Paper.pdf
_version_ 1779157186988998656