Title: Discover the GellMann-Okubo formula with machine learning

Abstract: Machine learning is a novel and powerful technology and has been widely used in various science topics. We demonstrate a machine-learning based approach built by a set of general metrics and rules inspired by physics. Taking advantages of physical constraints, such as dimension identity, symmetry and generalization, we succeed to rediscover the GellMann Okubo formula using a technique of symbolic regression. This approach can effectively find explicit solutions among user-defined observable, and easily extend to study on exotic hadron spectrum.