Skip to main content

Stepwise Evolutionary Learning using Deep Learned Guidance Functions

Johnson, Colin G. (2019) Stepwise Evolutionary Learning using Deep Learned Guidance Functions. In: Bramer, Max and Petridis, Miltos, eds. Lecture Notes in Artificial Intelligence. Artificial Intelligence XXXVI: 39th SGAI International Conference on Artificial Intelligence, AI 2019, Cambridge, UK, December 17–19, 2019, Proceedings. Lecture Notes in Computer Science , 11927. Springer ISBN 978-3-030-34884-7. (doi:10.1007/978-3-030-34885-4_4) (Access to this publication is currently restricted. You may be able to access a copy if URLs are provided) (KAR id:78198)

PDF Author's Accepted Manuscript
Language: English

Restricted to Repository staff only until 18 November 2020.
Contact us about this Publication
[img]
Official URL
https://doi.org/10.1007/978-3-030-34885-4_4

Abstract

This paper explores how Learned Guidance Functions (LGFs)— a pre-training method used to smooth search landscapes—can be used as a fitness function for evolutionary algorithms. A new form of LGF is introduced, based on deep neural network learning, and it is shown how this can be used as a fitness function. This is applied to a test problem: unscrambling the Rubik’s Cube. Comparisons are made with a previous LGF approach based on random forests, and with a baseline approach based on traditional error-based fitness.

Item Type: Conference or workshop item (Proceeding)
DOI/Identification number: 10.1007/978-3-030-34885-4_4
Uncontrolled keywords: evolutionary algorithms; Learned Guidance Functions
Subjects: Q Science > QA Mathematics (inc Computing science) > QA 76 Software, computer programming, > QA76.87 Neural computers, neural networks
Divisions: Faculties > Sciences > School of Computing
Depositing User: Colin Johnson
Date Deposited: 03 Nov 2019 22:41 UTC
Last Modified: 22 Jan 2020 16:11 UTC
Resource URI: https://kar.kent.ac.uk/id/eprint/78198 (The current URI for this page, for reference purposes)
Johnson, Colin G.: https://orcid.org/0000-0002-9236-6581
  • Depositors only (login required):

Downloads

Downloads per month over past year