论文部分内容阅读
利用共生进化原理设计人工神经网络 ,创造性地融入了免疫调节原理中的浓度抑制调节机制以保持个体的多样性 ,提出了基于免疫调节的共生进化网络设计方法 .通过对神经元群体而不是神经网络群体进行进化设计 ,显著地减轻了计算量 ,同时利用生物免疫原理中的浓度机制和个体多样性保持策略进行免疫调节 ,有效地克服了未成熟收敛现象 ,提高了群体的多样性 ,从而加快优化设计速度 .实验结果表明该方法可高效、准确地设计鲁棒性很强的神经网络 .
Using artificial symbiosis theory to design artificial neural network and creatively integrate concentration inhibitory mechanism in immune regulation principle to maintain individual diversity, a new symbiosis evolutionary network design method based on immunomodulation is proposed.By comparing neuronal population rather than neural network The evolutionary design of the population significantly reduces the computational load, and at the same time, it uses the mechanism of concentration in biological immunization principles and individual diversity maintenance strategies to immunomodulate, effectively overcoming immature convergence, increasing population diversity and accelerating optimization Design speed.The experimental results show that this method can design a robust neural network efficiently and accurately.