Learning Multiple Conflicting Tasks with Artificial Evolution

Research output: Contribution in Book/Catalog/Report/Conference proceedingChapter (peer-reviewed)

50 Downloads (Pure)

Abstract

The paper explores the issue of learning multiple competing tasks in the domain of artificial evolution. In particular, a robot is trained so as to be able to perform two different tasks at the same time, namely a gradient following and rough terrain avoidance behaviours. It is shown that, if the controller is trained to learn two tasks of different difficulty, then the robot performance is higher if the most difficult task is learnt first, before the combined learning of both tasks. An explanation to this superiority is also discussed, in comparison with previous results.

Original languageEnglish
Title of host publicationAdvances in Artificial Life and Evolutionary Computation
Subtitle of host publication9th Italian Workshop, WIVACE 2014, Vietri sul Mare, Italy, May 14-15, Revised Selected Papers
EditorsClara Pizzuti, Giandomenico Spezzano
PublisherSpringer
Pages127-139
Number of pages13
Volume445
ISBN (Electronic)978-3-319-12745-3
ISBN (Print)978-3-319-12744-6
DOIs
Publication statusPublished - 6 Nov 2014
Event9th Italian Workshop on Artificial Life and Evolutionary Computation, WIVACE 2014 - Vietri sul Mare, Italy
Duration: 14 May 201415 May 2014

Publication series

NameCommunications in Computer and Information Science
Volume445
ISSN (Print)18650929

Conference

Conference9th Italian Workshop on Artificial Life and Evolutionary Computation, WIVACE 2014
CountryItaly
CityVietri sul Mare
Period14/05/1415/05/14

Fingerprint

Robots
Controllers

Keywords

  • Artificial Intelligence
  • neural networks
  • evolutionary robotics
  • genetic algorithms

Cite this

Nicolay, D., Andrea, R., & Carletti, T. (2014). Learning Multiple Conflicting Tasks with Artificial Evolution. In C. Pizzuti, & G. Spezzano (Eds.), Advances in Artificial Life and Evolutionary Computation: 9th Italian Workshop, WIVACE 2014, Vietri sul Mare, Italy, May 14-15, Revised Selected Papers (Vol. 445, pp. 127-139). (Communications in Computer and Information Science; Vol. 445). Springer. https://doi.org/10.1007/978-3-319-12745-3_11
Nicolay, Delphine ; Andrea, Roli ; Carletti, Timoteo. / Learning Multiple Conflicting Tasks with Artificial Evolution. Advances in Artificial Life and Evolutionary Computation: 9th Italian Workshop, WIVACE 2014, Vietri sul Mare, Italy, May 14-15, Revised Selected Papers. editor / Clara Pizzuti ; Giandomenico Spezzano. Vol. 445 Springer, 2014. pp. 127-139 (Communications in Computer and Information Science).
@inbook{5fd703fb3313455482aa3a4bb38cdda4,
title = "Learning Multiple Conflicting Tasks with Artificial Evolution",
abstract = "The paper explores the issue of learning multiple competing tasks in the domain of artificial evolution. In particular, a robot is trained so as to be able to perform two different tasks at the same time, namely a gradient following and rough terrain avoidance behaviours. It is shown that, if the controller is trained to learn two tasks of different difficulty, then the robot performance is higher if the most difficult task is learnt first, before the combined learning of both tasks. An explanation to this superiority is also discussed, in comparison with previous results.",
keywords = "Artificial Intelligence, neural networks, evolutionary robotics, genetic algorithms",
author = "Delphine Nicolay and Roli Andrea and Timoteo Carletti",
year = "2014",
month = "11",
day = "6",
doi = "10.1007/978-3-319-12745-3_11",
language = "English",
isbn = "978-3-319-12744-6",
volume = "445",
series = "Communications in Computer and Information Science",
publisher = "Springer",
pages = "127--139",
editor = "Clara Pizzuti and Giandomenico Spezzano",
booktitle = "Advances in Artificial Life and Evolutionary Computation",

}

Nicolay, D, Andrea, R & Carletti, T 2014, Learning Multiple Conflicting Tasks with Artificial Evolution. in C Pizzuti & G Spezzano (eds), Advances in Artificial Life and Evolutionary Computation: 9th Italian Workshop, WIVACE 2014, Vietri sul Mare, Italy, May 14-15, Revised Selected Papers. vol. 445, Communications in Computer and Information Science, vol. 445, Springer, pp. 127-139, 9th Italian Workshop on Artificial Life and Evolutionary Computation, WIVACE 2014, Vietri sul Mare, Italy, 14/05/14. https://doi.org/10.1007/978-3-319-12745-3_11

Learning Multiple Conflicting Tasks with Artificial Evolution. / Nicolay, Delphine; Andrea, Roli; Carletti, Timoteo.

Advances in Artificial Life and Evolutionary Computation: 9th Italian Workshop, WIVACE 2014, Vietri sul Mare, Italy, May 14-15, Revised Selected Papers. ed. / Clara Pizzuti; Giandomenico Spezzano. Vol. 445 Springer, 2014. p. 127-139 (Communications in Computer and Information Science; Vol. 445).

Research output: Contribution in Book/Catalog/Report/Conference proceedingChapter (peer-reviewed)

TY - CHAP

T1 - Learning Multiple Conflicting Tasks with Artificial Evolution

AU - Nicolay, Delphine

AU - Andrea, Roli

AU - Carletti, Timoteo

PY - 2014/11/6

Y1 - 2014/11/6

N2 - The paper explores the issue of learning multiple competing tasks in the domain of artificial evolution. In particular, a robot is trained so as to be able to perform two different tasks at the same time, namely a gradient following and rough terrain avoidance behaviours. It is shown that, if the controller is trained to learn two tasks of different difficulty, then the robot performance is higher if the most difficult task is learnt first, before the combined learning of both tasks. An explanation to this superiority is also discussed, in comparison with previous results.

AB - The paper explores the issue of learning multiple competing tasks in the domain of artificial evolution. In particular, a robot is trained so as to be able to perform two different tasks at the same time, namely a gradient following and rough terrain avoidance behaviours. It is shown that, if the controller is trained to learn two tasks of different difficulty, then the robot performance is higher if the most difficult task is learnt first, before the combined learning of both tasks. An explanation to this superiority is also discussed, in comparison with previous results.

KW - Artificial Intelligence

KW - neural networks

KW - evolutionary robotics

KW - genetic algorithms

U2 - 10.1007/978-3-319-12745-3_11

DO - 10.1007/978-3-319-12745-3_11

M3 - Chapter (peer-reviewed)

AN - SCOPUS:84916593100

SN - 978-3-319-12744-6

VL - 445

T3 - Communications in Computer and Information Science

SP - 127

EP - 139

BT - Advances in Artificial Life and Evolutionary Computation

A2 - Pizzuti, Clara

A2 - Spezzano, Giandomenico

PB - Springer

ER -

Nicolay D, Andrea R, Carletti T. Learning Multiple Conflicting Tasks with Artificial Evolution. In Pizzuti C, Spezzano G, editors, Advances in Artificial Life and Evolutionary Computation: 9th Italian Workshop, WIVACE 2014, Vietri sul Mare, Italy, May 14-15, Revised Selected Papers. Vol. 445. Springer. 2014. p. 127-139. (Communications in Computer and Information Science). https://doi.org/10.1007/978-3-319-12745-3_11