BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic Programming

We introduce BERT mutation, a novel, domain-independent mutation operator for Genetic Programming (GP) that leverages advanced Natural Language Processing (NLP) techniques to improve convergence, particularly using the Masked Language Modeling approach. By combining the capabilities of deep reinforc...

Full description

Saved in:
Bibliographic Details
Main Authors: Eliad Shem-Tov, Moshe Sipper, Achiya Elyasaf
Format: Article
Language:English
Published: MDPI AG 2025-02-01
Series:Mathematics
Subjects:
Online Access:https://www.mdpi.com/2227-7390/13/5/779
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850228195832365056
author Eliad Shem-Tov
Moshe Sipper
Achiya Elyasaf
author_facet Eliad Shem-Tov
Moshe Sipper
Achiya Elyasaf
author_sort Eliad Shem-Tov
collection DOAJ
description We introduce BERT mutation, a novel, domain-independent mutation operator for Genetic Programming (GP) that leverages advanced Natural Language Processing (NLP) techniques to improve convergence, particularly using the Masked Language Modeling approach. By combining the capabilities of deep reinforcement learning and the BERT transformer architecture, BERT mutation intelligently suggests node replacements within GP trees to enhance their fitness. Unlike traditional stochastic mutation methods, BERT mutation adapts dynamically by using historical fitness data to optimize mutation decisions, resulting in more effective evolutionary improvements. Through comprehensive evaluations across three benchmark domains, we demonstrate that BERT mutation significantly outperforms conventional and state-of-the-art mutation operators in terms of convergence speed and solution quality. This work represents a pivotal step toward integrating state-of-the-art deep learning into evolutionary algorithms, pushing the boundaries of adaptive optimization in GP.
format Article
id doaj-art-06586f2740244f76907d2b4db8dd2b93
institution OA Journals
issn 2227-7390
language English
publishDate 2025-02-01
publisher MDPI AG
record_format Article
series Mathematics
spelling doaj-art-06586f2740244f76907d2b4db8dd2b932025-08-20T02:04:36ZengMDPI AGMathematics2227-73902025-02-0113577910.3390/math13050779BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic ProgrammingEliad Shem-Tov0Moshe Sipper1Achiya Elyasaf2Department of Software and Information Systems Engineering, Ben-Gurion University of the Negev, Beer-Sheva 8410501, IsraelDepartment of Computer Science, Ben-Gurion University, Beer-Sheva 8410501, IsraelDepartment of Software and Information Systems Engineering, Ben-Gurion University of the Negev, Beer-Sheva 8410501, IsraelWe introduce BERT mutation, a novel, domain-independent mutation operator for Genetic Programming (GP) that leverages advanced Natural Language Processing (NLP) techniques to improve convergence, particularly using the Masked Language Modeling approach. By combining the capabilities of deep reinforcement learning and the BERT transformer architecture, BERT mutation intelligently suggests node replacements within GP trees to enhance their fitness. Unlike traditional stochastic mutation methods, BERT mutation adapts dynamically by using historical fitness data to optimize mutation decisions, resulting in more effective evolutionary improvements. Through comprehensive evaluations across three benchmark domains, we demonstrate that BERT mutation significantly outperforms conventional and state-of-the-art mutation operators in terms of convergence speed and solution quality. This work represents a pivotal step toward integrating state-of-the-art deep learning into evolutionary algorithms, pushing the boundaries of adaptive optimization in GP.https://www.mdpi.com/2227-7390/13/5/779genetic programmingmutation operatorreinforcement learningcombinatorial optimizationsurrogate modelsymbolic regression
spellingShingle Eliad Shem-Tov
Moshe Sipper
Achiya Elyasaf
BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic Programming
Mathematics
genetic programming
mutation operator
reinforcement learning
combinatorial optimization
surrogate model
symbolic regression
title BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic Programming
title_full BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic Programming
title_fullStr BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic Programming
title_full_unstemmed BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic Programming
title_short BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic Programming
title_sort bert mutation deep transformer model for masked uniform mutation in genetic programming
topic genetic programming
mutation operator
reinforcement learning
combinatorial optimization
surrogate model
symbolic regression
url https://www.mdpi.com/2227-7390/13/5/779
work_keys_str_mv AT eliadshemtov bertmutationdeeptransformermodelformaskeduniformmutationingeneticprogramming
AT moshesipper bertmutationdeeptransformermodelformaskeduniformmutationingeneticprogramming
AT achiyaelyasaf bertmutationdeeptransformermodelformaskeduniformmutationingeneticprogramming