Visual and Audio Monitoring of Island Based Parallel Evolutionary Algorithms

Monitoring and visualisation tools are currently attracting more and more attention in order to understand how search spaces are explored by complex optimisation ecosystems such as parallel evolutionary algorithms based on island models. Multilevel visualisation is actually a desirable feature for f...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of grid computing Ročník 13; číslo 3; s. 309 - 327
Hlavní autoři: Lutton, Evelyne, Gilbert, Hugo, Cancino, Waldo, Bach, Benjamin, Pallamidessi, Joseph, Parrend, Pierre, Collet, Pierre
Médium: Journal Article
Jazyk:angličtina
Vydáno: Dordrecht Springer Netherlands 01.09.2015
Springer Nature B.V
Springer Verlag
Témata:
Computer Science
Computer simulation
Environmental monitoring
Evolutionary algorithms
Genetic algorithms
Islands
Levers
Life Sciences
Management of Computing and Information Systems
Monitoring
Multilevel
Optimization
Processor Architectures
User Interfaces and Human Computer Interaction
Visual
Visualization
Audio monitoring
Island model
Musicalisation
Computational ecosystem
Visualisation
EASEA
Parallel evolutionary algorithms
ISSN:1570-7873, 1572-9184
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Monitoring and visualisation tools are currently attracting more and more attention in order to understand how search spaces are explored by complex optimisation ecosystems such as parallel evolutionary algorithms based on island models. Multilevel visualisation is actually a desirable feature for facilitating the monitoring of computationally expensive runs involving several hundreds of computers during hours or even days. In this paper we present two components of a future multilevel monitoring system: MusEAc, a high level, audio monitoring allowing to listen to a run and tune it in real time and GridVis, a lower lever, more precise a posteriori visualisation tool that lets the user understand why the algorithm has performed well or bad.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:1570-7873
1572-9184
DOI:10.1007/s10723-014-9321-8