Open Questions For Suprathreshold Stochastic Resonance In Sensory Neural Models for Motion Detection Using Artificial Insect Vision

Abstract

Abstract. Stochastic Resonance (SR) occurs when the presence of noise in a nonlinear system can induce an optimal output from that system, and has been observed in a diverse range of physical and biological systems, including neurons. Despite this widespread observation of SR, to date very few engineering applications inspired by SR have been proposed, and one of the goals of our research is to explore possible new practical applications designed to replicate the benefits of SR. In particular, since about 1991, our group has designed and implemented a number of motion detection VLSI chips based on insect vision. We are currently investigating the possibility of replicating the benefits of SR in artificial insect-vision based motion detection systems, in particular a newly described form of SR called Suprathreshold Stochastic Resonance (SSR). The current paper is intended to review and identify the key open questions and avenues for future research relating to SR and SSR in such systems.

Noise-Improved Bayesian Estimation With Arrays of One-Bit Quantizers

“…In addition, in various circumstances, the analog input may not be accessible when no linear transducers are available for this. It will be the case, for instance, with sensory neurons, which form a thresholded all-or-nothing type of representation of an analog signal from the physical environment, bearing similarities with the representation by the present array of comparators [27], [29], [37]. This will also be the case with nanodevices with intrinsic nonlinearities and usable in building sensor microarrays [38].…”