欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
基于提高神经网络泛化性能的目标提出了神经网络泛化损失率的概念，解析了与前一周期相比当前网络误差的变化趋势，在此基础上导出了基于泛化损失率的神经网络训练目标函数。利用新的目标函数和基于量子化粒子群算法的神经网络训练方法，得到了一种新的网络泛化性能优化算法。实验结果表明，将该算法与没有引入泛化损失率的算法相比，网络的收敛性能和泛化性能都有明显提高。
英文摘要：
For the purpose of improving the generalization ability of neural network, a novel concept of generalization loss rate is proposed. The concept represents the growing trends of network output error compared with the previous period. The objective function of network training based on it is also proposed. By making use of the new objective fimction and the network training method based on quantum-behaved particle swarm optimization, a novel algorithm to improve the network generalization performance is achieved. Ex- perimental result show that the new algorithm has better generalize and convergent ability compared with those who do not import the generalization loss rate.
同期刊论文项目
同项目期刊论文
期刊信息
位置：