Lexicase Selection Outperforms Previous Strategies for Incremental Evolution of Virtual Creature Controllers

Jared M. Moore; Adam Stanton

doi:10.1162/isal_a_050

Lexicase Selection Outperforms Previous Strategies for Incremental Evolution of Virtual Creature Controllers

Jared M. Moore, Adam Stanton

Computer Science

Research output: Chapter in Book/Published conference output › Conference publication

Abstract

Evolving robust behaviors for robots has proven to be a challenging problem. Determining how to optimize behavior for a specific instance, while also realizing behaviors that generalize to variations on the problem often requires highly customized algorithms and problem-specific tuning of the evolutionary platform. Algorithms that can realize robust, generalized behavior without this customization are therefore highly desirable. In this paper, we examine the Lexicase selection algorithm as a possible general algorithm for a wall crossing robot task. Previous work has resulted in specialized strategies to evolve robust behaviors for this task. Here, we show that Lexicase selection is not only competitive with these strategies but after parameter tuning, actually exceeds the performance of the specialized algorithms.

Original language	English
Title of host publication	Proceedings of ECAL 2017
Editors	Carole Knibbe, others
Pages	290-297
DOIs	https://doi.org/10.1162/isal_a_050
Publication status	Published - 1 Sep 2017
Event	ECAL 2017, the Fourteenth European Conference on Artificial Life - Lyons, France Duration: 4 Sep 2017 → 8 Sep 2017

Conference

Conference	ECAL 2017, the Fourteenth European Conference on Artificial Life
Abbreviated title	ECAL 2017
Country/Territory	France
City	Lyons
Period	4/09/17 → 8/09/17

Bibliographical note

©2017 Massachusetts Institute of Technology. This work is licensed to the public under a Creative Commons Attribution - NonCommercial - NoDerivatives 4.0 license (international): https://creativecommons.org/licenses/by-nc-nd/4.0/

Access to Document

10.1162/isal_a_050

isal_a_050
©2017 Massachusetts Institute of Technology). This work is licensed to the public under a Creative Commons Attribution - NonCommercial - NoDerivatives 4.0 license (international): https://creativecommons.org/licenses/by-nc-nd/4.0/
Final published version, 1.5 MBLicence: CC BY-NC-ND 4.0

Cite this

@inproceedings{4f502fd21d674358bf864d5362bdce9b,

title = "Lexicase Selection Outperforms Previous Strategies for Incremental Evolution of Virtual Creature Controllers",

abstract = "Evolving robust behaviors for robots has proven to be a challenging problem. Determining how to optimize behavior for a specific instance, while also realizing behaviors that generalize to variations on the problem often requires highly customized algorithms and problem-specific tuning of the evolutionary platform. Algorithms that can realize robust, generalized behavior without this customization are therefore highly desirable. In this paper, we examine the Lexicase selection algorithm as a possible general algorithm for a wall crossing robot task. Previous work has resulted in specialized strategies to evolve robust behaviors for this task. Here, we show that Lexicase selection is not only competitive with these strategies but after parameter tuning, actually exceeds the performance of the specialized algorithms. ",

author = "Moore, {Jared M.} and Adam Stanton",

note = "{\textcopyright}2017 Massachusetts Institute of Technology. This work is licensed to the public under a Creative Commons Attribution - NonCommercial - NoDerivatives 4.0 license (international): https://creativecommons.org/licenses/by-nc-nd/4.0/; ECAL 2017, the Fourteenth European Conference on Artificial Life, ECAL 2017 ; Conference date: 04-09-2017 Through 08-09-2017",

year = "2017",

month = sep,

day = "1",

doi = "10.1162/isal_a_050",

language = "English",

isbn = "978-0-262-34633-7",

pages = "290--297",

editor = "Carole Knibbe and others",

booktitle = "Proceedings of ECAL 2017",

}

TY - GEN

T1 - Lexicase Selection Outperforms Previous Strategies for Incremental Evolution of Virtual Creature Controllers

AU - Moore, Jared M.

AU - Stanton, Adam

N1 - ©2017 Massachusetts Institute of Technology. This work is licensed to the public under a Creative Commons Attribution - NonCommercial - NoDerivatives 4.0 license (international): https://creativecommons.org/licenses/by-nc-nd/4.0/

PY - 2017/9/1

Y1 - 2017/9/1

N2 - Evolving robust behaviors for robots has proven to be a challenging problem. Determining how to optimize behavior for a specific instance, while also realizing behaviors that generalize to variations on the problem often requires highly customized algorithms and problem-specific tuning of the evolutionary platform. Algorithms that can realize robust, generalized behavior without this customization are therefore highly desirable. In this paper, we examine the Lexicase selection algorithm as a possible general algorithm for a wall crossing robot task. Previous work has resulted in specialized strategies to evolve robust behaviors for this task. Here, we show that Lexicase selection is not only competitive with these strategies but after parameter tuning, actually exceeds the performance of the specialized algorithms.

AB - Evolving robust behaviors for robots has proven to be a challenging problem. Determining how to optimize behavior for a specific instance, while also realizing behaviors that generalize to variations on the problem often requires highly customized algorithms and problem-specific tuning of the evolutionary platform. Algorithms that can realize robust, generalized behavior without this customization are therefore highly desirable. In this paper, we examine the Lexicase selection algorithm as a possible general algorithm for a wall crossing robot task. Previous work has resulted in specialized strategies to evolve robust behaviors for this task. Here, we show that Lexicase selection is not only competitive with these strategies but after parameter tuning, actually exceeds the performance of the specialized algorithms.

UR - https://machinemachines.org/wp-content/uploads/2018/06/moore-stanton-ecal-2017.pdf

UR - https://direct.mit.edu/isal/proceedings/ecal2017/290/99553

U2 - 10.1162/isal_a_050

DO - 10.1162/isal_a_050

M3 - Conference publication

SN - 978-0-262-34633-7

SP - 290

EP - 297

BT - Proceedings of ECAL 2017

A2 - Knibbe, Carole

A2 - others,

T2 - ECAL 2017, the Fourteenth European Conference on Artificial Life

Y2 - 4 September 2017 through 8 September 2017

ER -

Lexicase Selection Outperforms Previous Strategies for Incremental Evolution of Virtual Creature Controllers

Abstract

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this