Geometric Deep Learning for the Rubik’s Cube Group
- 21 August 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 36 (12) , 20214-20228
Abstract
The Rubik’s cube, a widely recognized combinatorial puzzle with an astronomically vast state space, has been the subject of various research experiments with neural networks used as heuristic estimators to navigate the state-space exploration. However, prior efforts have overlooked the intriguing symmetries inherent to this domain. Drawing on geometric deep learning principles, this article introduces a novel neural architecture that explicitly leverages these symmetries, grounded in a rigorous group-theoretical analysis. The design of the proposed symmetry-invariant model is then validated empirically through an innovative universal procedure for detecting model symmetry invariance. Finally, experimental results demonstrate that the symmetry-aware neural architecture exhibits enhanced generalization and problem-solving efficacy compared with the state of the art.Keywords
Funding Information
- Grantová Agentura České Republiky (24-11664S)
This publication has 21 references indexed in Scilit:
- Towards efficiently solving the rubik’s cube with deep reinforcement learning and recursionE3S Web of Conferences, 2024
- Geometric deep learning for drug discoveryExpert Systems with Applications, 2023
- Commutators in the Rubik’s Cube GroupThe American Mathematical Monthly, 2023
- Casting Rubik’s Group into a Unitary Representation for Reinforcement LearningJournal of Physics: Conference Series, 2023
- Symmetry and complexity in object-centric deep active inference modelsInterface Focus, 2023
- Geometric deep learning on molecular representationsNature Machine Intelligence, 2021
- Solving Rubik’s cube via quantum mechanics and deep reinforcement learningJournal of Physics A: Mathematical and Theoretical, 2021
- Highly accurate protein structure prediction with AlphaFoldNature, 2021
- Stepwise Evolutionary Learning Using Deep Learned Guidance Functions2019
- Solving the Rubik’s cube with deep reinforcement learning and searchNature Machine Intelligence, 2019