基于忆阻器的连续学习混沌神经网络  

Memristor-based Successive Learning Chaotic Neural Network

作　　者：张椅[1] 段书凯[1] 王丽丹[1] 胡小方[1,2] 

机构地区：[1]西南大学物理科学与技术学院电子信息工程学院,重庆400715 [2]香港城市大学机械与生物医学工程系

出　　处：《计算机科学》2013年第10期213-217,共5页Computer Science

基　　金：国家自然科学基金(60972155;61101233;60974020);中央高校基本科研业务费专项资金(XDJK2012A007);重庆市高等学校青年骨干教师资助计划(渝教人〔2011〕65号);重庆市高等学校优秀人才支持计划(渝教人〔2011〕65号);教育部"春晖计划"科研项目(z2011148);留学人员科技活动项目择优资助经费(国家级优秀类;渝人社办[2012]186号;102060-20600601)资助

摘　　要：忆阻器具有独特的记忆功能和连续可变的电导状态,在人工智能与神经网络等研究领域具有巨大的应用优势。详细推导了忆阻器的电荷控制模型,将纳米忆阻器与具有智能信息处理能力的混沌神经网络相结合,提出了一种新型的基于忆阻器的连续学习混沌神经网络模型。利用忆阻器可直接实现网络中繁多的反馈与迭代,即完成外部输入对神经元及神经元之间相互作用的时空总和。提出的忆阻连续学习混沌神经网络可以实现对已知模式和未知模式的区分,并能对未知模式进行自动学习和记忆。给出的计算机仿真验证了方案的可行性。由于忆阻器具有纳米级尺寸和自动的记忆能力,该方案有望大大简化混沌神经网络结构。With the unique memory ability and continuously variable conductance state, memristors have promising prospects in the fields of artificial intelligence and artificial neural network. This paper derived the charge-controlled memristor model in detail. Combining the nanometer memristor and chaotic neural network, a novel type of memristor-based successive learning chaotic neural network model was proposed. The numerous feedbacks and iterative in the net- work, that is, the spaticrtemporal summation of external input to neurons and the interaction between neurons, can been achieved by taking advantage of memristor. In the proposed model, it makes use of the difference in the response to the input patterns to distinguish the unknown pattern from the stored known patterns. When an input pattern is regarded as an unknown pattern,it will be memorized in the network. The effectiveness was verified through the given simulation experiments. With the memristor^s nano-scale size and automatic memory capacity, the program is expected to greatly simplify the structure of chaotic neural network.

关 键 词：忆阻器 混沌神经网络 连续学习 时空总和 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]