Fitness Importance for online evolution

Philip Valencia; Raja Jurdak; Peter Lindsay

doi:10.1145/1830761.1830890

Fitness Importance for online evolution

Philip Valencia^*, Raja Jurdak, Peter Lindsay

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

Abstract

To complement standard fitness functions, we propose "Fitness Importance" (FI) as a novel meta-heuristic for online learning systems. We define FI and show how it can be used to dynamically bias the population composition in order to vary the instantaneous system performance at a tradeoff to learning capability. The effect of FI is demonstrated on a simple light-sensing and light-actuating optimisation problem running on multiple wireless sensor network devices. We also describe how FI can be used with the In situ Distributed Genetic Programming (IDGP) framework to balance learning and performing for resource-constrained computing devices which evolve their logic continuously.

Original language	English
Title of host publication	Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10 - Companion Publication
Place of Publication	New York
Publisher	ACM
Pages	2117-2118
Number of pages	2
ISBN (Print)	9781450300735
DOIs	https://doi.org/10.1145/1830761.1830890
Publication status	Published - 2010
Externally published	Yes
Event	12th Annual Genetic and Evolutionary Computation Conference, GECCO-2010 - Portland, OR, United States Duration: 7 Jul 2010 → 11 Jul 2010

Other

Other	12th Annual Genetic and Evolutionary Computation Conference, GECCO-2010
Country/Territory	United States
City	Portland, OR
Period	7/07/10 → 11/07/10

Keywords

Adaptive
Evolution
Fitness
Genetic Program
Late breaking abstract
Objective
Online
Wireless sensor network

Access to Document

10.1145/1830761.1830890

Cite this

@inproceedings{28db489c399149c98fa87de7da0583ed,

title = "Fitness Importance for online evolution",

abstract = "To complement standard fitness functions, we propose {"}Fitness Importance{"} (FI) as a novel meta-heuristic for online learning systems. We define FI and show how it can be used to dynamically bias the population composition in order to vary the instantaneous system performance at a tradeoff to learning capability. The effect of FI is demonstrated on a simple light-sensing and light-actuating optimisation problem running on multiple wireless sensor network devices. We also describe how FI can be used with the In situ Distributed Genetic Programming (IDGP) framework to balance learning and performing for resource-constrained computing devices which evolve their logic continuously.",

keywords = "Adaptive, Evolution, Fitness, Genetic Program, Late breaking abstract, Objective, Online, Wireless sensor network",

author = "Philip Valencia and Raja Jurdak and Peter Lindsay",

year = "2010",

doi = "10.1145/1830761.1830890",

language = "English",

isbn = "9781450300735",

pages = "2117--2118",

booktitle = "Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10 - Companion Publication",

publisher = "ACM",

note = "12th Annual Genetic and Evolutionary Computation Conference, GECCO-2010 ; Conference date: 07-07-2010 Through 11-07-2010",

}

Valencia, P, Jurdak, R & Lindsay, P 2010, Fitness Importance for online evolution. in Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10 - Companion Publication. ACM, New York, pp. 2117-2118, 12th Annual Genetic and Evolutionary Computation Conference, GECCO-2010, Portland, OR, United States, 7/07/10. https://doi.org/10.1145/1830761.1830890

TY - GEN

T1 - Fitness Importance for online evolution

AU - Valencia, Philip

AU - Jurdak, Raja

AU - Lindsay, Peter

PY - 2010

Y1 - 2010

N2 - To complement standard fitness functions, we propose "Fitness Importance" (FI) as a novel meta-heuristic for online learning systems. We define FI and show how it can be used to dynamically bias the population composition in order to vary the instantaneous system performance at a tradeoff to learning capability. The effect of FI is demonstrated on a simple light-sensing and light-actuating optimisation problem running on multiple wireless sensor network devices. We also describe how FI can be used with the In situ Distributed Genetic Programming (IDGP) framework to balance learning and performing for resource-constrained computing devices which evolve their logic continuously.

AB - To complement standard fitness functions, we propose "Fitness Importance" (FI) as a novel meta-heuristic for online learning systems. We define FI and show how it can be used to dynamically bias the population composition in order to vary the instantaneous system performance at a tradeoff to learning capability. The effect of FI is demonstrated on a simple light-sensing and light-actuating optimisation problem running on multiple wireless sensor network devices. We also describe how FI can be used with the In situ Distributed Genetic Programming (IDGP) framework to balance learning and performing for resource-constrained computing devices which evolve their logic continuously.

KW - Adaptive

KW - Evolution

KW - Fitness

KW - Genetic Program

KW - Late breaking abstract

KW - Objective

KW - Online

KW - Wireless sensor network

UR - http://www.scopus.com/inward/record.url?scp=77955950559&partnerID=8YFLogxK

U2 - 10.1145/1830761.1830890

DO - 10.1145/1830761.1830890

M3 - Conference proceeding contribution

AN - SCOPUS:77955950559

SN - 9781450300735

SP - 2117

EP - 2118

BT - Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10 - Companion Publication

PB - ACM

CY - New York

T2 - 12th Annual Genetic and Evolutionary Computation Conference, GECCO-2010

Y2 - 7 July 2010 through 11 July 2010

ER -

Fitness Importance for online evolution

Abstract

Other

Keywords

Access to Document

Other files and links

Fingerprint

Cite this