Novel Spike based Reservoir Node Design with High Performance Spike Delay Loop
Chenyuan Zhao, Jialing Li, Lingjia Liu, Lakshmi Sravanthi Koutha, Jian Liu, Yang Yi · 2016
With the emerging cutting-edge semiconductor nanotechnologies, reservoir computing has shifted the focus from software implementation towards hardware and optical implementations over the last few decades. Nowadays, in the field of reservoir computing, pure analog implementation with the employment of CMOS nanotechnology has attracted a worldwide attention. In this paper, we introduce a spike-based analog reservoir node with compact delay performing at extremely high accuracy. The detailed work on this novel spike processor working in near chaotic condition is presented and analyzed. Furthermore, we announce a novel class of spike-based delay loop which enables the pure spike implementation of reservoir computing. The simulation results demonstrated that our delay loop possesses exceptionally high accuracy of 99.24% and gain-hold property without adopting any kind of amplifiers. Our proposed reservoir node has shown the potential of transforming spike signals to a high-dimensional state space.