Mathematical expression exploration with graph representation and generative graph neural network.

Journal: Neural networks : the official journal of the International Neural Network Society
PMID:

Abstract

Symbolic Regression (SR) methods in tree representations have exhibited commendable outcomes across Genetic Programming (GP) and deep learning search paradigms. Nonetheless, the tree representation of mathematical expressions occasionally embodies redundant substructures. Representing expressions as computation graphs is more succinct and intuitive through graph representation. Despite its adoption in evolutionary strategies within SR, deep learning paradigms remain under-explored. Acknowledging the profound advancements of deep learning in tree-centric SR approaches, we advocate for addressing SR tasks using the Directed Acyclic Graph (DAG) representation of mathematical expressions, complemented by a generative graph neural network. We name the proposed method as Graph-based Deep Symbolic Regression (GraphDSR). We vectorize node types and employ an adjacent matrix to delineate connections. The graph neural networks craft the DAG incrementally, sampling node types and graph connections conditioned on previous DAG at every step. During each sample step, the valid check is implemented to avoid meaningless sampling, and four domain-agnostic constraints are adopted to further streamline the search. This process culminates once a coherent expression emerges. Constants undergo optimization by SGD and BFGS algorithms, and rewards refine the graph neural network through reinforcement learning. A comprehensive evaluation across 110 benchmarks underscores the potency of our approach.

Authors

  • Jingyi Liu
    College of Sciences, Northeastern University, Shenyang 110819, China.
  • Weijun Li
    Institute of Semiconductors, Chinese Academy of Sciences, Beijing, 100083, China; School of Integrated Circuits, University of Chinese Academy of Sciences, Beijing, 100083, China; Beijing Key Laboratory of Semiconductor Neural Network Intelligent Sensing and Computing Technology, Beijing, 100083, China. Electronic address: wjli@semi.ac.cn.
  • Lina Yu
    Institute of Semiconductors, Chinese Academy of Sciences, Beijing, 100083, China.
  • Min Wu
    Guizhou University of Traditional Chinese Medicine, Guiyang, Guizhou Province, China.
  • Wenqiang Li
    Department of Interventional Neuroradiology, Beijing Neurosurgical Institute and Beijing Tiantan Hospital, Capital Medical University, No.119 South Fourth Ring West Road, Fengtai District, Beijing, 100050, China.
  • Yanjie Li
    College of Science, Shenyang Aerospace University, 110136, People's Republic of China.
  • Meilan Hao
    AnnLab, Institute of Semiconductors, Chinese Academy of Sciences, Beijing, 100083, China; School of Integrated Circuits & Center of Materials Science and Optoelectronics Engineering, University of Chinese Academy of Sciences, 100049, Beijing, China. Electronic address: mlhao@semi.ac.cn.