【摘要】 针对情感神经网络（emotional neural network,ENN）出现的预测精度不足和泛化能力有限问题，提出一种基于免疫调节和细胞机制的自适应情感神经网络。借鉴免疫系统调节机制，构建一种自适应的免疫调节ENN(adaptive-immunomodulation ENN,AIENN)。在隐含层结构设计中，采用高斯和升余弦双激活函数，增强网络的非线性表达能力。此外，根据生物细胞分裂和凋亡机制对网络结构大小动态调整，提高网络泛化能力。采用分数阶梯度下降法进行参数学习，提高网络的预测精度。仿真实验结果表明，所提AIENN和其它神经网络算法相比，具有更高的预测精度和更好的自适应能力。更多还原

【Abstract】 To solve the problems of insufficient prediction accuracy and limited generalization ability in the emotional neural network(ENN), an adaptive emotional neural network based on immune regulation and cellular mechanisms was proposed. An adaptive immunomodulation ENN(AIENN) was constructed by drawing on the mechanism of immune system regulation. In the design of the hidden layer structure, Gaussian and raised cosine double activation functions were used to enhance the nonlinear expression ability of the network. The size of the network structure was dynamically adjusted according to the mechanism of biological cell division and apoptosis, and the network’s generalization ability was improved. Fractional gradient descent was employed for parameter learning, thereby enhancing the prediction accuracy of the network. Simulation results show that the proposed AIENN achieves higher prediction accuracy and better adaptability compared to other neural network algorithms.更多还原