PCA Landscape Projections for Meta-Analysis of Genetic Improvement for Source Code

Created by W.Langdon from gp-bibliography.bib Revision:1.9159

@InProceedings{Nemeth:2026:UKCI,

• author = "Zsolt Nemeth and Penn {Faulkner Rainford} and Barry Porter",
• title = "{PCA} Landscape Projections for Meta-Analysis of Genetic Improvement for Source Code",
• booktitle = "25th UK Workshop on Computational Intelligence (UKCI 2026)",
• year = "2026",
• address = "Coventry University",
• month = "9--11 " # sep,
• keywords = "genetic algorithms, genetic programming, Genetic Improvement, Fitness Landscapes, Principal Component Analysis, Performance Optimisation, hash experiments, Kullback-Leibler fitness",
• URL = "http://www.cs.ucl.ac.uk/staff/W.Langdon/ftp/papers/Nemeth_2026_UKCI.pdf",
• size = "12 pages",
• abstract = "Genetic improvement (GI) for source code uses evolutionary methods to improve either the functional or non-functional elements of existing software. We consider GI for performance improvement, which has notably different requirements to traditional bug-fixing applications of GI: in performance improvement we mix together genetic programming (synthesis of new material) with genetic improvement, and the extent to which the genotype and phenotype landscapes are aligned (together with the mutation grammar) determines the success of a GI search. Because GI process are inherently complex and have a wide range of tunable parameters, gaining a clear understanding of which GI mechanisms lead to which outcomes is of crucial importance in identifying successful GI mechanisms for a given problem type. We introduce a PCA-based landscape analysis method to gain a detailed understanding of what happens during a GI execution, both at the level of genetic individuals and of entire experiments under various configurations",
• notes = "https://ukci2026.coventry.ac.uk/",

}

Genetic Programming entries for Zsolt Nemeth Penelope Faulkner Rainford Barry Porter

Citations