RL-GEP: Symbolic Regression via Gene Expression Programming and Reinforcement Learning

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

@InProceedings{conf/ijcnn/ZhangZ21a,

• author = "Hengzhe Zhang and Aimin Zhou",
• title = "{RL-GEP}: Symbolic Regression via Gene Expression Programming and Reinforcement Learning",
• booktitle = "2021 International Joint Conference on Neural Networks, IJCNN",
• year = "2021",
• address = "Shenzhen, China",
• month = "18-22 " # jul,
• publisher = "IEEE",
• keywords = "genetic algorithms, genetic programming, gene expression programming, reinforcement learning, symbolic regression",
• isbn13 = "978-1-6654-3900-8",
• bibdate = "2021-09-29",
• bibsource = "DBLP, http://dblp.uni-trier.de/db/conf/ijcnn/ijcnn2021.html#ZhangZ21a",
• URL = "https://ieeexplore.ieee.org/document/9533735",
• DOI = "doi:10.1109/IJCNN52387.2021.9533735",
• size = "8 pages",
• abstract = "Symbolic regression has become a hot topic in recent years due to the surging demand for interpretable machine learning methods. Traditionally, symbolic regression problems are mainly solved by genetic algorithms. Nonetheless, with the development of deep learning, reinforcement learning based symbolic regression methods have received attention gradually. Unfortunately, hardly any of those reinforcement learning based methods have been proven effectively to solve real world regression problems as genetic algorithm based methods. In this paper, we find a general reinforcement learning based symbolic regression method is difficult to solve real world problems since it is hard to balance between exploration and exploitation. To deal with this problem, we propose a hybrid method to use both genetic algorithm and reinforcement learning for solving symbolic regression problems. By doing so, we can combine the advantages of reinforcement learning and genetic algorithm and achieve better performance than using them alone. To validate the effectiveness of the proposed method, we apply the proposed method to ten benchmark datasets. The experimental results show that the proposed method achieves competitive performance compared with several well-known symbolic regression methods on those datasets.",
• notes = "Institute of AI Education, School of Computer Science and Technology East China Normal University, Shanghai 200062, China",

}

Genetic Programming entries for Hengzhe Zhang Aimin Zhou

Citations