• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Combinatorial Optimizer Based on Deep Reinforcement Learning

Research Project

Project/Area Number 20K11988
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 61040:Soft computing-related
Research InstitutionKyoto Institute of Technology

Principal Investigator

IIMA Hitoshi  京都工芸繊維大学, 情報工学・人間科学系, 准教授 (70273547)

Project Period (FY) 2020-04-01 – 2023-03-31
Project Status Completed (Fiscal Year 2022)
Budget Amount *help
¥4,420,000 (Direct Cost: ¥3,400,000、Indirect Cost: ¥1,020,000)
Fiscal Year 2022: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2021: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2020: ¥2,080,000 (Direct Cost: ¥1,600,000、Indirect Cost: ¥480,000)
Keywords機械学習 / 深層学習 / 強化学習 / 深層強化学習 / 最適化 / 組合せ最適化 / モンテカルロ木探索
Outline of Research at the Start

組合せ最適化はありとあらゆる分野、例えば生産、物流、通信ネットワーク、金融、土木、農業などの分野で用いられているが、多くの最適化問題は難解であり、短時間で最適な解を求める方法が存在しない。一方、囲碁やビデオゲームに対して、人間のプレイヤーより上手にプレイする機械学習技術が提案され、非常に注目を集めている。この機械学習は強化学習を深層学習と組み合わせたもので、深層強化学習と呼ばれている。本研究では、この深層強化学習の学習能力を活かした独創的な組合せ最適化法を開発する。また、機械学習の汎化能力を活かして、学習の大半を最初に一度だけ実行するだけで、種々の問題に対する解を短時間に得る方法を開発する。

Outline of Final Research Achievements

We have developed optimizers based on deep reinforcement learning for solving combinatorial optimization problems, which are known to be challenging to find optimal solutions in a short time. These optimizers are classified into two types: Go and video games. We have also studied the basic framework of a method that can find a solution for a new instance in a short time after pre-training for another instance. These methods were applied to a delivery scheduling problem, and their performance was investigated experimentally.

Academic Significance and Societal Importance of the Research Achievements

深層強化学習が学習問題に対して高い性能を示すことが知られているが、これを組合せ最適化法として構成し直すことにより、従来の方法とは一線を画す解法を開発できた。この解法を既存の最適化法と比較することにより、さらに優れた最適化法を開発することが期待できる。また、優れた最適化法を開発することで、生産、物流、通信ネットワーク、金融、交通、土木、農業などの多くの分野のシステム開発に寄与できる。

Report

(4 results)
  • 2022 Annual Research Report   Final Research Report ( PDF )
  • 2021 Research-status Report
  • 2020 Research-status Report
  • Research Products

    (9 results)

All 2022 2021 2020

All Journal Article (6 results) (of which Peer Reviewed: 6 results,  Open Access: 1 results) Presentation (3 results) (of which Int'l Joint Research: 3 results)

  • [Journal Article] Generative Adversarial Network for Generating Different Types of Data2022

    • Author(s)
      Murota Shingo、Iima Hitoshi
    • Journal Title

      IEEJ Transactions on Electronics, Information and Systems

      Volume: 142 Issue: 7 Pages: 781-787

    • DOI

      10.1541/ieejeiss.142.781

    • ISSN
      0385-4221, 1348-8155
    • Year and Date
      2022-07-01
    • Related Report
      2022 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Drone Delivery Scheduling by a Genetic Algorithm: Simultaneous Optimization for Multiple Different Numbers of Drones2022

    • Author(s)
      狭間陽平,飯間 等,軽野義行
    • Journal Title

      IEEJ Transactions on Electronics, Information and Systems

      Volume: 142 Issue: 4 Pages: 499-505

    • DOI

      10.1541/ieejeiss.142.499

    • ISSN
      0385-4221, 1348-8155
    • Year and Date
      2022-04-01
    • Related Report
      2022 Annual Research Report
    • Peer Reviewed
  • [Journal Article] 報酬の設定を自動化した集中型高速マルチエージェント強化学習法2022

    • Author(s)
      佐々木薫、飯間 等
    • Journal Title

      システム制御情報学会論文誌

      Volume: 35 Pages: 39-47

    • Related Report
      2021 Research-status Report
    • Peer Reviewed
  • [Journal Article] Genetic algorithm for scheduling of parcel delivery by drones2021

    • Author(s)
      Yohei Hazama, Hitoshi Iima, Yoshiyuki Karuno and Kosuke Mishima
    • Journal Title

      Journal of Advanced Mechanical Design, Systems, and Manufacturing

      Volume: 15 Issue: 6 Pages: JAMDSM0069-JAMDSM0069

    • DOI

      10.1299/jamdsm.2021jamdsm0069

    • NAID

      130008104555

    • ISSN
      1881-3054
    • Related Report
      2021 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Monte Carlo Tree Search Method for Solving the Knapsack Problem2020

    • Author(s)
      飯間 等、兵野拓海
    • Journal Title

      IEEJ Transactions on Electronics, Information and Systems

      Volume: 140 Issue: 10 Pages: 1141-1146

    • DOI

      10.1541/ieejeiss.140.1141

    • NAID

      130007920900

    • ISSN
      0385-4221, 1348-8155
    • Year and Date
      2020-10-01
    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] 経路探索問題に対して深層学習との併用により汎化能力を向上させた強化学習法2020

    • Author(s)
      飯間 等、大西鴻哉
    • Journal Title

      計測自動制御学会論文集

      Volume: 56 Issue: 10 Pages: 455-462

    • DOI

      10.9746/sicetr.56.455

    • NAID

      130007925078

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Presentation] Genetic Algorithm with Machine Learning to Estimate the Optimal Objective Function Values of Subproblems2022

    • Author(s)
      Hitoshi Iima, Yohei Hazama
    • Organizer
      6th International Conference on Intelligent Systems, Metaheuristics & Swarm Intelligence
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Hyperheuristic Method Based on Deep Reinforcement Learning2022

    • Author(s)
      Hitoshi Iima, Yoshiyuki Nakamura
    • Organizer
      11th International Congress on Advanced Applied Informatics
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Design of an Artificial Game Entertainer by Reinforcement Learning2020

    • Author(s)
      Takanobu Yaguchi, Hitoshi Iima
    • Organizer
      IEEE Conference on Games
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research

URL: 

Published: 2020-04-28   Modified: 2024-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi