Population Dynamics in Genetic Programming for Dynamic Symbolic Regression

This paper investigates the application of genetic programming (GP) for dynamic symbolic regression (SR), addressing the challenge of adapting machine learning models to evolving data in practical applications. Benchmark instances with changing underlying functions over time are defined to assess th...

Full description

Saved in:
Bibliographic Details
Published in:Applied sciences Vol. 14; no. 2; p. 596
Main Authors: Fleck, Philipp, Werth, Bernhard, Affenzeller, Michael
Format: Journal Article
Language:English
Published: Basel MDPI AG 01.01.2024
Subjects:
Adaptation
Algorithms
Artificial intelligence
dynamic optimization
Genetic algorithms
genetic programming
Genetic research
Interoperability
Machine learning
Neural networks
Optimization
Population biology
Programming languages
symbolic regression
Austria
Germany
ISSN:2076-3417, 2076-3417
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper investigates the application of genetic programming (GP) for dynamic symbolic regression (SR), addressing the challenge of adapting machine learning models to evolving data in practical applications. Benchmark instances with changing underlying functions over time are defined to assess the performance of a genetic algorithm (GA) as a traditional evolutionary algorithm and an age-layered population structure (ALPS) as an open-ended evolutionary algorithm for dynamic symbolic regression. This study analyzes population dynamics by examining variable frequencies and impact changes over time in response to dynamic shifts in the training data. The results demonstrate the effectiveness of both the GA and ALPS in handling changing data, showcasing their ability to recover and evolve improved solutions after an initial drop in population quality following data changes. Population dynamics reveal that variable impacts respond rapidly to data changes, while variable frequencies shift gradually across generations, aligning with the indirect measure of fitness represented by variable impacts. Notably, the GA shows a strong dependence on mutation to avoid variables becoming permanently extinct, contrasting with the ALPS’s unexpected insensitivity to mutation rates owing to its reseeding mechanism for effective variable reintroduction.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2076-3417
2076-3417
DOI:10.3390/app14020596