New pulse amplitude modulation for fine tuning of memristor synapses

详细信息    查看全文

• 作者：Son Ngoc Truong^a ; Khoa Van Pham^a ; Wonsun Yang^a ; Sangho Shin^b ; Ken Pedrotti^c ; Kyeong-Sik Min^a ; ^{mks@kookmin.ac.kr" class="auth_mail" title="E-mail the corresponding author}
• 关键词：Fine tuning of memristor synapses ; Pulse amplitude modulation ; Cellular neural networks ; Neuromorphic circuits
• 刊名：Microelectronics Journal
• 出版年：2016
• 出版时间：September 2016
• 年：2016
• 卷：55
• 期：Complete
• 页码：162-168
• 全文大小：2569 K

文摘

Nanoscale memristors can be used as synapses in brain-mimicking neuromorphic systems. Here, the fine tuning of memristor conductance is important in controlling the synapse weights precisely, because the coarse tuning of memristor synapses can cause a significant error in neuromorphic processing. In this paper, we propose a new Pulse Amplitude Modulation (PAM) method for the fine tuning of memristor conductance. The new PAM scheme is verified by the experimental measurement of real memristors, where the new PAM could reduce the pulse-to-pulse fluctuation in conductance change per pulse by 84.8%, compared to the previous linear PAM. For comparing the linear and new PAM schemes, they are tested in programming memristor synapses in the memristor-based Cellular Neural Networks (CNN). The simulation result confirms that the new-PAM-programmed CNN shows better quality of edge detection than the linear-PAM-programmed CNN.