忆阻器是具有记忆和类突触特性的非线性电路元件，将忆阻器与STDP学习规则相结合，提出了基于STDP学习规则的忆阻神经网络，并将它应用于二值图像的叠加和灰度图像的存储与输出。首先将忆阻器作突触，通过实验证实在特定形状动作电位下，可实现STDP学习规则；构建了4×4的忆阻交叉阵列神经网络；用16×16的忆阻交叉阵列神经网络实现二值图像的叠加。最后用N×N的忆阻神经网络实现了灰度图像的存储与输出。通过MATLAB仿真实验证实了该方案的有效性，该忆阻神经网络具有仿生特性，有望解决模式识别、人工智能中出现的复杂问题。%Memristor is a nonlinear circuit element which has property of memory and the similar synapse characteristic. In this paper, a memristive neural network based on STDP learning rule is proposed and used in the binary image overlay and the storage and output of gray and color images. Firstly, the memristor is used as the synapse, and the specific shape action potentials realizing STDP learning rules are demonstrated. Secondly, 4×4 memristive crossbar array neural networks are realized. And then the binary image overlay is obtained by the neural network of 16×16 memristive crossbar array neural networks. Finally, the storage and output of gray and color images are achieved by the N×N memristive neural network. The feasibility of the proposed method is proved by MATLAB simulation experiments. This neural network has more bionic feasibility, which could be applied to solve the complex problems in pattern recognition and artificial intelligence.
展开▼
