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

2017 Fiscal Year Annual Research Report

Research on Large-Scale Algorithm Portfolios

Research Project

Project/Area Number 25330253
Research InstitutionThe University of Tokyo

Principal Investigator

福永 ALEX  東京大学, 大学院総合文化研究科, 教授 (90452002)

Project Period (FY) 2013-04-01 – 2018-03-31
Keywords人工知能 / 探索アルゴリズム / 並列探索アルゴリズム
Outline of Annual Research Achievements

We continued our investigation of the reduction of communications overhead among processors in a parallel best-first search algorithm.
Previously in this project, we developed Abstract Zobrist Hashing (AZH) for balancing communications overhead and workload distribution. AZH requires a projection from the actual search space into an abstract search space, but our previous methods required ad hoc methods for projecting the search space into an abstract search space.
This year, we developed GRAZHDA*, a general approach to completely automatically generating the feature projection functions used by AZH based on graph partitioning.
GRAZHDA* seeks to approximate the partitioning of the search space (and hence the workload distribution) by partitioning the domain transition graph, an abstraction of the state space graph. GRAZHDA* subsumes the previously projection methods introduced in previous work and can be automatically applied to domains expressed in PDDL, a standard language for expressing domain-independent planning problems. We showed that GRAZHDA* using a sparsity-based objective function for partitioning successfully automatically extracts projection functions which enable AZH to achieve a good balance between load balance and communications overhead.

  • Research Products

    (2 results)

All 2017

All Journal Article (2 results) (of which Peer Reviewed: 2 results,  Open Access: 1 results)

  • [Journal Article] On Hash-Based Work Distribution Methods for Parallel Best-First Search2017

    • Author(s)
      Yuu Jinnai, Alex fukunaga
    • Journal Title

      Journal of Artificial Intelligence Research

      Volume: 60 Pages: 491-548

    • DOI

      https://doi.org/10.1613/jair.5225

    • Peer Reviewed / Open Access
  • [Journal Article] TPAM: a simulation-based model for quantitatively analyzing parameter adaptation methods2017

    • Author(s)
      Ryoji Tanabe, Alex Fukunaga
    • Journal Title

      Proceedings of the Genetic and Evolutionary Computation Conference

      Volume: 1 Pages: 729-736

    • Peer Reviewed

URL: 

Published: 2018-12-17  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi