Neurogenetic programming framework for explainable reinforcement learning

Vadim Liventsev; Aki Härmä; Milan Petkovic

doi:10.1145/3449726.3459537

Neurogenetic programming framework for explainable reinforcement learning

Vadim Liventsev
, Aki Härmä
, Milan Petkovic

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

Abstract

Automatic programming, the task of generating computer programs compliant with a specification without a human developer, is usually tackled either via genetic programming methods based on mutation and recombination of programs, or via neural language models. We propose a novel method that combines both approaches using a concept of a virtual neuro-genetic programmer, or scrum team. We demonstrate its ability to provide performant and explainable solutions for various OpenAI Gym tasks, as well as inject expert knowledge into the otherwise data-driven search for solutions.

Original language	English
Title of host publication	GECCO 2021 Companion - Proceedings of the 2021 Genetic and Evolutionary Computation Conference Companion
Publisher	The Association for Computing Machinery, Inc.
Pages	329-330
Number of pages	2
ISBN (Electronic)	9781450383516
DOIs	https://doi.org/10.1145/3449726.3459537
Publication status	Published - 7 Jul 2021
Externally published	Yes
Event	2021 Genetic and Evolutionary Computation Conference - Online, Lille, France Duration: 10 Jul 2021 → 14 Jul 2021 https://gecco-2021.sigevo.org/HomePage

Conference

Conference	2021 Genetic and Evolutionary Computation Conference
Abbreviated title	GECCO 2021
Country/Territory	France
City	Lille
Period	10/07/21 → 14/07/21
Internet address	https://gecco-2021.sigevo.org/HomePage

Keywords

genetic programming
program synthesis
reinforcement learning

Access to Document

10.1145/3449726.3459537Licence: Other

Cite this

@inproceedings{3adc46c8c80940c18ff3c85523909d93,

title = "Neurogenetic programming framework for explainable reinforcement learning",

abstract = "Automatic programming, the task of generating computer programs compliant with a specification without a human developer, is usually tackled either via genetic programming methods based on mutation and recombination of programs, or via neural language models. We propose a novel method that combines both approaches using a concept of a virtual neuro-genetic programmer, or scrum team. We demonstrate its ability to provide performant and explainable solutions for various OpenAI Gym tasks, as well as inject expert knowledge into the otherwise data-driven search for solutions.",

keywords = "genetic programming, program synthesis, reinforcement learning",

author = "Vadim Liventsev and Aki H{\"a}rm{\"a} and Milan Petkovic",

note = "Funding Information: This work was funded by the European Union{\textquoteright}s Horizon 2020 research and innovation programme under grant agreement 812882. This work is part of {"}Personal Health Interfaces Leveraging HUman-MAchine Natural interactionS{"} (PhilHumans) project Publisher Copyright: {\textcopyright} 2021 Owner/Author.; 2021 Genetic and Evolutionary Computation Conference, GECCO 2021 ; Conference date: 10-07-2021 Through 14-07-2021",

year = "2021",

month = jul,

day = "7",

doi = "10.1145/3449726.3459537",

language = "English",

pages = "329--330",

booktitle = "GECCO 2021 Companion - Proceedings of the 2021 Genetic and Evolutionary Computation Conference Companion",

publisher = "The Association for Computing Machinery, Inc.",

url = "https://gecco-2021.sigevo.org/HomePage",

}

Liventsev, V, Härmä, A & Petkovic, M 2021, Neurogenetic programming framework for explainable reinforcement learning. in GECCO 2021 Companion - Proceedings of the 2021 Genetic and Evolutionary Computation Conference Companion. The Association for Computing Machinery, Inc., pp. 329-330, 2021 Genetic and Evolutionary Computation Conference, Lille, France, 10/07/21. https://doi.org/10.1145/3449726.3459537

Neurogenetic programming framework for explainable reinforcement learning. / Liventsev, Vadim; Härmä, Aki; Petkovic, Milan.
GECCO 2021 Companion - Proceedings of the 2021 Genetic and Evolutionary Computation Conference Companion. The Association for Computing Machinery, Inc., 2021. p. 329-330.

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

TY - GEN

T1 - Neurogenetic programming framework for explainable reinforcement learning

AU - Liventsev, Vadim

AU - Härmä, Aki

AU - Petkovic, Milan

N1 - Funding Information: This work was funded by the European Union’s Horizon 2020 research and innovation programme under grant agreement 812882. This work is part of "Personal Health Interfaces Leveraging HUman-MAchine Natural interactionS" (PhilHumans) project Publisher Copyright: © 2021 Owner/Author.

PY - 2021/7/7

Y1 - 2021/7/7

N2 - Automatic programming, the task of generating computer programs compliant with a specification without a human developer, is usually tackled either via genetic programming methods based on mutation and recombination of programs, or via neural language models. We propose a novel method that combines both approaches using a concept of a virtual neuro-genetic programmer, or scrum team. We demonstrate its ability to provide performant and explainable solutions for various OpenAI Gym tasks, as well as inject expert knowledge into the otherwise data-driven search for solutions.

AB - Automatic programming, the task of generating computer programs compliant with a specification without a human developer, is usually tackled either via genetic programming methods based on mutation and recombination of programs, or via neural language models. We propose a novel method that combines both approaches using a concept of a virtual neuro-genetic programmer, or scrum team. We demonstrate its ability to provide performant and explainable solutions for various OpenAI Gym tasks, as well as inject expert knowledge into the otherwise data-driven search for solutions.

KW - genetic programming

KW - program synthesis

KW - reinforcement learning

UR - https://www.scopus.com/pages/publications/85111016357

U2 - 10.1145/3449726.3459537

DO - 10.1145/3449726.3459537

M3 - Conference article in proceeding

AN - SCOPUS:85111016357

SP - 329

EP - 330

BT - GECCO 2021 Companion - Proceedings of the 2021 Genetic and Evolutionary Computation Conference Companion

PB - The Association for Computing Machinery, Inc.

T2 - 2021 Genetic and Evolutionary Computation Conference

Y2 - 10 July 2021 through 14 July 2021

ER -

Neurogenetic programming framework for explainable reinforcement learning

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this