DOI: 10.1109/IJCNN.2017.7966423
关键词: Equilibrium point 、 Artificial neural network 、 Applied mathematics 、 Mathematical optimization 、 Exponential stability 、 Recurrent neural network 、 Generalization 、 Uniqueness 、 Matrix (mathematics) 、 Mathematics 、 Quaternion
摘要: This paper introduces matrix-valued recurrent neural networks with time delays, and proves the existence uniqueness of global equilibrium point. These are a generalization complex-, quaternion- Clifford-valued matrix states. Two sufficient criteria derived in terms linear inequalities that ensure asymptotic stability point for proposed networks. Finally, two simulation examples demonstrate effectiveness theoretical results.