论文部分内容阅读
本文将原连续时间电压域点格神经网络（CNN）模型[1]映射到离散时间电流域，并采用开关电流（SI）技术来进行离散时间CNN的VLSI设计。给出了基本积木块的实现电路，并进行了连通片检测方面的简单应用研究。计算机仿真与理论结果相吻合，所得电路非常适合于用标准CMOS工艺集成，具有潜在的应用前景。
In this paper, the original continuous-time voltage-domain lattice neural network (CNN) model [1] is mapped to the discrete-time current domain, and the switched-current (SI) technique is used to design the VLSI of discrete-time CNN. The realization of the basic building block is given, and a simple application research on the detection of connected pieces is given. Computer simulation and theoretical results are consistent, the resulting circuit is very suitable for integration with standard CMOS technology, has potential applications.