Searching Search Spaces: Meta-evolving a Geometric Encoding for Neural Networks

Created by W.Langdon from gp-bibliography.bib Revision:1.8051

@InProceedings{kunze:2024:CEC,

• author = "Tarek Kunze and Paul Templier and Dennis G. Wilson",
• title = "Searching Search Spaces: Meta-evolving a Geometric Encoding for Neural Networks",
• booktitle = "2024 IEEE Congress on Evolutionary Computation (CEC)",
• year = "2024",
• editor = "Bing Xue",
• address = "Yokohama, Japan",
• month = "30 " # jun # " - 5 " # jul,
• publisher = "IEEE",
• keywords = "genetic algorithms, genetic programming, cartesian genetic programming, Neurons, Genomics, Evolutionary computation, Encoding, Bioinformatics, Biological neural networks, evolution strategies, meta-evolution, neural networks, reinforcement learning, policy search",
• isbn13 = "979-8-3503-0837-2",
• DOI = "doi:10.1109/CEC60901.2024.10612026",
• abstract = "In evolutionary policy search, neural networks are usually represented using a direct mapping: each gene encodes one network weight. Indirect encoding methods, where each gene can encode for multiple weights, shorten the genome to reduce the dimensions of the search space and better exploit permutations and symmetries. The Geometric Encoding for Neural network Evolution (GENE) introduced an indirect encoding where the weight of a connection is computed as the (pseudo-)distance between the two linked neurons, leading to a genome size growing linearly with the number of genes instead of quadratically in direct encoding. However GENE still relies on hand -crafted distance functions with no prior optimisation. Here we show that better performing distance functions can be found for GENE using Cartesian Genetic Programming (CGP) in a meta-evolution approach, hence optimising the encoding to create a search space that is easier to exploit. We show that GENE with a learnt function can outperform both direct encoding and the hand-crafted distances, generalizing on unseen problems, and we study how the encoding impacts neural network properties.",
• notes = "also known as \cite{10612026}

  WCCI 2024",

}

Genetic Programming entries for Tarek Kunze Paul Templier Dennis G Wilson

Citations