Parametric Genetic Programming

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

@InProceedings{Pindur:2020:SCIS-ISIS,

• author = "Adam Kotaro Pindur and Takahiro Horiba and Hitoshi Iba",
• title = "Parametric Genetic Programming",
• booktitle = "2020 Joint 11th International Conference on Soft Computing and Intelligent Systems and 21st International Symposium on Advanced Intelligent Systems (SCIS-ISIS)",
• year = "2020",
• abstract = "In conventional Genetic Programming for regression problems, the whole population evolves to adapt to the training data. However, individuals cannot adjust themselves to provided data, which results in slow convergence. On the other hand, gradient-based methods such as gradient descent quickly converge toward the solution by using local gradient information. Local search methods can be applied in Genetic Programming at an individual level, thus introducing local adaptations. In this paper, we propose to use the inner product of the input data with the parameter vector, which is updated using the Gradient Descent method. This approach is beneficial for two reasons: (i) gradient descent optimizes models created by the evolutionary process, thus decreases the search space of the GP, and (ii) inner product reduces the dimension of the input data, thus allows GP to efficiently handle high-dimensional data. Additionally, we propose an efficient algorithm to obtain information on the gradient. The proposed method is tested for regression problems on datasets with 10 or more dimensions. Results show that the proposed method converges to solutions faster than the conventional GP, at the same time providing models comparable to models created by SVM.",
• keywords = "genetic algorithms, genetic programming, Standards, Gradient methods, Mathematical model, Convergence, Training, Taylor series, Stochastic Gradient Descent, Symbolic Regression",
• DOI = "doi:10.1109/SCISISIS50064.2020.9322682",
• month = dec,
• notes = "Also known as \cite{9322682}",

}

Genetic Programming entries for Adam Kotaro Pindur Takahiro Horiba Hitoshi Iba

Citations