量子自组织特征映射神经网络OA

Self-organizing feature mapping is a typical unsupervised neural network algorithm.It adopts competitive learning strategies to achieve data classification.However,when the number of neurons in the network is polynomial the training of self-organized feature mapping algorithm will be challenged by computational power.In order to reduce the time complexity of the algorithm,a quantum-classical hybrid self-organizing feature mapping neural network model is proposed.It provides a parallel scheme for similarity calculation and label extraction by using quantum superposition and quantum entanglement during the training process of neural network.Theoretical analysis shows that the proposed algorithm has exponential acceleration in data dimension compared with the classical algorithm.