A Genetic Programming Encoder for Increasing Autoencoder Interpretability

Created by W.Langdon from gp-bibliography.bib Revision:1.7642

@InProceedings{Schofield:2023:EuroGP,

• author = "Finn Schofield and Luis Slyfield and Andrew Lensen",
• title = "A Genetic Programming Encoder for Increasing Autoencoder Interpretability",
• booktitle = "EuroGP 2023: Proceedings of the 26th European Conference on Genetic Programming",
• year = "2023",
• month = "12-14 " # apr,
• editor = "Gisele Pappa and Mario Giacobini and Zdenek Vasicek",
• series = "LNCS",
• volume = "13986",
• publisher = "Springer Verlag",
• address = "Brno, Czech Republic",
• pages = "19--35",
• organisation = "EvoStar, Species",
• keywords = "genetic algorithms, genetic programming, Autoencoder, Dimensionality reduction, Machine learning, Explainable artificial intelligence, XAI",
• isbn13 = "978-3-031-29572-0",
• URL = "https://openaccess.wgtn.ac.nz/ndownloader/files/39034739",
• URL = "https://rdcu.be/c8UN0",
• DOI = "doi:10.1007/978-3-031-29573-7_2",
• size = "17 pages",
• abstract = "Autoencoders are powerful models for non-linear dimensionality reduction. However, their neural network structure makes it difficult to interpret how the high dimensional features relate to the low-dimensional embedding, which is an issue in applications where explainability is important. There have been attempts to replace both the neural network components in autoencoders with interpretable genetic programming (GP) models. However, for the purposes of interpretable dimensionality reduction, we observe that replacing only the encoder with GP is sufficient. In this work, we propose the Genetic Programming Encoder for Autoencoding (GPE-AE). GPE-AE uses a multi-tree GP individual as an encoder, while retaining the neural network decoder. We demonstrate that GPE-AE is a competitive non-linear dimensionality reduction technique compared to conventional autoencoders and a GP based method that does not use an autoencoder structure. As visualisation is a common goal for dimensionality reduction, we also evaluate the quality of visualisations produced by our method, and highlight the value of functional mappings by demonstrating insights that can be gained from interpreting the GP encoders.",
• notes = "Part of \cite{Pappa:2023:GP} EuroGP'2023 held in conjunction with EvoCOP2023, EvoMusArt2023 and EvoApplications2023",

}

Genetic Programming entries for Finn Schofield Luis Slyfield Andrew Lensen

Citations