Decoupling Representation and Learning in Genetic Programming: the LaSER Approach

Created by W.Langdon from gp-bibliography.bib Revision:1.8414

@Misc{le2025decouplingrepresentationlearninggenetic,

• author = "Nam H. Le and Josh Bongard",
• title = "Decoupling Representation and Learning in Genetic Programming: the LaSER Approach",
• howpublished = "arXiv 2505.17309",
• year = "2025",
• month = "6 " # jun,
• edition = "v2",
• keywords = "genetic algorithms, genetic programming, symbolic tree GP, DEAP, LASER, Baldwin Effect, lifetime learning: Linear Regression, Ridge Regression, scikit-learn, Decision Tree, Random Forest, Gradient Boosting, Multi-layer Perceptron, MLP, ANN",
• archiveprefix = "arXiv",
• primaryclass = "cs.NE",
• URL = "https://arxiv.org/abs/2505.17309",
• size = "15 pages",
• abstract = "Genetic Programming (GP) has traditionally entangled the evolution of symbolic representations with their performance-based evaluation, often relying solely on raw fitness scores. This tight coupling makes GP solutions more fragile and prone to overfitting, reducing their ability to generalize. we propose LaSER (Latent Semantic Representation Regression) -- a general framework that decouples representation evolution from lifetime learning. At each generation, candidate programs produce features which are passed to an external learner to model the target task. This approach enables any function approximator, from linear models to neural networks, to serve as a lifetime learner, allowing expressive modeling beyond conventional symbolic forms.

  Here we show for the first time that LaSER can out compete standard GP and GP followed by linear regression when it employs non-linear methods to fit coefficients to GP-generated equations against complex data sets. Further, we explore how LaSER enables the emergence of innate representations, supporting long-standing hypotheses in evolutionary learning such as the Baldwin Effect. By separating the roles of representation and adaptation, LaSER offers a principled and extensible framework for symbolic regression and classification.",

• notes = "'LaSER evolves programs that act as feature generators' cites \cite{keijzer03} 'semantic vector supports downstream learning'

  Population size: 100, Generations: 100, Function set: {+, -, *, /, sin, cos, exp, log}, 700 training examples.

  Learning is non-inheritable.

  Nguyen benchmark functions + Vladislavleva Suite.

  perfectly innate, potential Baldwinian assimilation.

  See also \cite{le:2025:GPTP}",

}

Genetic Programming entries for Nam Le Josh C Bongard

Citations