HIGH-LEVEL SEARCH Instead of directly solving a High-level Search - - PowerPoint PPT Presentation

▶
high level search
SMART_READER_LITE
LIVE PREVIEW

HIGH-LEVEL SEARCH Instead of directly solving a High-level Search - - PowerPoint PPT Presentation

Apr 19, 2023 288 likes •533 views

HIGH-LEVEL SEARCH: FROM HYPER- HEURISTICS TO ALGORITHM SELECTION MUSTAFA MISIR HIGH-LEVEL SEARCH Instead of directly solving a High-level Search Methods given probem (~instance), Operates upon perform a search on algorithm space

slide-1
SLIDE 1

HIGH-LEVEL SEARCH: FROM HYPER- HEURISTICS TO ALGORITHM SELECTION MUSTAFA MISIR

slide-2
SLIDE 2

HIGH-LEVEL SEARCH

  • Instead of directly solving a

given probem (~instance), perform a search on algorithm space

  • Generic (Problem-

Independent) Methods

High-level Search Methods Low-level Algorithms Potential Solutions

Operates upon Operates upon

slide-3
SLIDE 3

HIGH-LEVEL SEARCH

  • Can be performed:
  • ONLINE: while solving a problem instance
  • OFFLINE: before solving a problem instance
slide-4
SLIDE 4

ONLINE

slide-5
SLIDE 5

A GENERIC INTELLIGENT HYPER-HEURISTIC

Mustafa MISIR Patrick De CAUSMAECKER Greet VANDEN BERGHE Katja VERBEECK

slide-6
SLIDE 6

OUTLINE

  • OBJECTIVE
  • HYPER-HEURISTICS
  • A GENERIC INTELLIGENT HH (GIHH)
  • COMPUTATIONAL RESULTS
slide-7
SLIDE 7

OUTLINE

  • OBJECTIVE
  • HYPER-HEURISTICS
  • A GENERIC INTELLIGENT HH (GIHH)
  • COMPUTATIONAL RESULTS
slide-8
SLIDE 8

OBJECTIVE

Designing a generic problem solver for solving all kinds of search and optimisation problems under different conditions

slide-9
SLIDE 9

OUTLINE

  • OBJECTIVE
  • HYPER-HEURISTICS
  • A GENERIC INTELLIGENT HH (GIHH)
  • COMPUTATIONAL RESULTS
slide-10
SLIDE 10

HYPER-HEURISTICS | DEFINITION

Burke et al. (2009): “A hyper-heuristic is a search method

  • r

learning mechanism for selecting

  • r

generating heuristics to solve computational search problems”

slide-11
SLIDE 11

OUTLINE

  • OBJECTIVE
  • HYPER-HEURISTICS
  • A GENERIC INTELLIGENT HH (GIHH)
  • COMPUTATIONAL RESULTS
slide-12
SLIDE 12
slide-13
SLIDE 13
slide-14
SLIDE 14

ADAPTIVE DYNAMIC HEURISTIC SET

  • Use an elite subset of a given heuristic set
  • During different parts of a search process (based on

the heuristics’ performance)

slide-15
SLIDE 15
slide-16
SLIDE 16
slide-17
SLIDE 17
slide-18
SLIDE 18

MOVE ACCEPTANCE + RE-INITIALISATION

  • Adaptive iterion

limited list-based threshold accepting

  • If doesn’t work,

re-initialise

slide-19
SLIDE 19

OUTLINE

  • OBJECTIVE
  • HYPER-HEURISTICS
  • A GENERIC INTELLIGENT HH (GIHH)
  • COMPUTATIONAL RESULTS
slide-20
SLIDE 20

COMPUTATIONAL RESULTS

  • HyFlex: Hyper-heuristic

software framework (Uni. Nottingham)

  • 6 problem domains
  • Different algorithm

(heuristic) sets

  • GIHH Open source @

http://allserv.kahosl.be/~mustafa.misir/gihh.html

slide-21
SLIDE 21

COMPUTATIONAL RESULTS | CHESC 2011

CHeSC 2011 WINNER !

Best algorithm on average across 6 problem domains with a large score margin Againist 19 competitors

slide-22
SLIDE 22

DISCUSSION

  • A generic high-level search tool
  • GIHH as the CHeSC 2011 Winner
  • Robust, easily applicable (generic)
  • Goal: combine offline and online
slide-23
SLIDE 23