The Application of Graph Neural Networks in Maze Problem
Mathematical Modeling and Algorithm Application · 15.03.2026 · 9 (1) · 403-407
DOI: 10.54097/gq83vf96 ↗Kurz gefasst
Fachartikel in der Zeitschrift Mathematical Modeling and Algorithm Application, online erschienen am 15.03.2026. Verfasst von Xinyang Chen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Graph Neural Network (GNN) shows significant applications in many fields these days, being a possible strong solution for Maze Problem, like finding paths and predicting distance. This paper attempts to apply Graph Neural Network with traditional algorithms as solutions. First, this paper built a generator to construct mazes as the data set for later experiments. Then, this paper proposes two methods. One is GNN-ShortestPath, that predicts distance to the target position by aggregating messages along the graph. The other one is GNN-Policy, which generates distribution of actions in steps. This paper performs experiments based on the methods and evaluates the results on Successful Rate and Optimal Rate. Finally, this paper analyses results and discusses limitation and possible improvements. Despite of many limitations in the experiments, this paper shows that application of Graph Neural Network in solving Maze Problem has great advantages and traditional algorithms can assist in the performance of Graph Neural Network.
Zitieren
APA
Chen, X. (2026). The Application of Graph Neural Networks in Maze Problem. Mathematical Modeling and Algorithm Application, 9(1), 403-407. https://doi.org/10.54097/gq83vf96
Vancouver
Chen X. The Application of Graph Neural Networks in Maze Problem. Mathematical Modeling and Algorithm Application. 2026;9(1):403-407. doi:10.54097/gq83vf96
BibTeX
@article{Chen2026,
author = {Chen, Xinyang},
title = {The Application of Graph Neural Networks in Maze Problem},
journal = {Mathematical Modeling and Algorithm Application},
year = {2026},
volume = {9},
number = {1},
pages = {403--407},
doi = {10.54097/gq83vf96}
}
EUVIMED