Review of Reinforcement Learning for Combinatorial Optimization Problem

The solution methods for combinatorial optimization problem (COP) have permeated to the fields of artificial intelligence, operations research, etc. With the scale of data increasing and the speed of problem updating being faster, the traditional method of solving the COP is challenged in computatio...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Jisuanji kexue yu tansuo Ročník 16; číslo 2; s. 261 - 279
Hlavný autor: WANG Yang, CHEN Zhibin, WU Zhaorui, GAO Yuan
Médium: Journal Article
Jazyk:Chinese
Vydavateľské údaje: Journal of Computer Engineering and Applications Beijing Co., Ltd., Science Press 01.02.2022
Predmet:
reinforcement learning (rl)|deep reinforcement learning (drl)|combinatorial optimization problem (cop)
ISSN:1673-9418
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:The solution methods for combinatorial optimization problem (COP) have permeated to the fields of artificial intelligence, operations research, etc. With the scale of data increasing and the speed of problem updating being faster, the traditional method of solving the COP is challenged in computational speed, precision and generali-zation ability. In recent years, reinforcement learning (RL) has been widely used in driverless, industrial automation and other fields, showing strong decision-making and learning ability. Thus, many researchers have strived to use RL to solve COP, which provides a novel method for solving these problems. This paper firstly introduces the common COP problems and the basic principles of RL. Then, this paper elaborates the difficulties of RL in solving COP, analyzes the advantages of RL in combinatorial optimization (CO) field, and studies the principle of the combina-tion of RL and COP. Subsequently, this paper summarizes the theoretical methods and applied researches of solving CO
ISSN:1673-9418
DOI:10.3778/j.issn.1673-9418.2107040