On the Positive Role of Noise and Cross-Talk of Elements in Neuromorphic Systems
Victor Erokhin
Roles of three important properties of the neuromorphic systems are considered in this paper: effect of noise, cross-talk of elements and energy efficiency. The positive effect of noise, directly demonstrated theoretically and experimentally during recent years, allows the system to reach a global energy level, providing effective classification and forecasting tasks execution. Cross-talk of elements is important for the realization of bio-realistic learning algorithms and mimicking cognitive and creative processes. Energy efficiency is an obvious important parameter for all modern systems. In this paper we underline the necessity of applying new hardware approaches for the effective solving of this problem.
Keywords: Neuromorphic systems, memristive devices, noise, cross-talk of elements, energy efficiency