摘要
首先介绍了在神经密码应用中的树型奇偶机模型,在综述神经密码协议研究的基础上,构建了一个使用含有衰减项的神经网络同步学习规则的新的密钥交换协议,并对其进行了理论分析。仿真实验结果表明,此协议可以使神经网络的同步性能提高4倍以上,对几种常见攻击的抵御表现良好,与传统的协议相比,新协议同步计算开销更小,安全性更高。
This paper first introduced the tree parity machine (TPM) network model. On the basis of this study, it proposed a novel neural network with attenuation term. Several simulations were then conducted to show that synchronization efficiency of the new protocol is improved more than 80%. Meanwhile, it also indicates that the new protocol is computational inexpen- sive and it improves the security of the protocol compared to the classic TPM-based protocol.
出处
《计算机应用研究》
CSCD
北大核心
2014年第10期3090-3092,3099,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(61170249)
重庆市自然科学基金面上项目(CSTC2010BB2210)
重庆市博士后研究人员科研基金特别资助项目(渝xm201103013)
关键词
神经网络同步
密钥交换
树型奇偶机
衰减项
neural synchronization
key exchange
tree parity machines(TPM)
attenuation term