Neuromorphic Systems: Past, Present and Future

Abstract

Neuromorphic systems are implementations in silicon of elements of neural systems. The idea of electronic implementation is not new, but modern microelectronics has provided opportunities for producing systems for both sensing and neural modelling that can be mass produced straightforwardly. We review the the history of neuromorphic systems, and discuss the range of neuromorphic systems that have ben developed. We discuss recent ideas for overcoming some of the problems, particularly providing effective adaptive synapses in large numbers.

Synchronization conditions in simple memristor neural networks

“…2 The convergence rate is named as c 2 to denote the fact that the contraction theorem makes explicit use of the Euclidean vector-normalso known as vector-2 norm and defined as ||v|| 2 = m j=1 |v j | 2 …”

“…Interestingly, the architecture of neuromorphic circuits [2], bio-inspired electronic systems which mimick the spatio-temporal pattern recognition processes and the neural signal processing mechanisms occurring in the brain of a living being, is based upon networks of nonlinear and dynamical cells.…”

The End of Moore’s Law: Opportunities for Natural Computing?

“…It is, therefore, no surprise that research to make hardware mimicking the brain has a long history (see for example [57] for a review). Rather than giving a comprehensive overview, we discuss the principal differences with conventional computers and the future potential of brain-inspired computing in this section.…”

Teaching Memory Circuit Elements via Experiment-Based Learning

“…For instance, it is presently accepted that biological synapses play the major role in brain memory and information processing. Analogously, in artificial neural networks [44], memristive devices can be used as artificial synapses [45], [46], [11], [10], [9], [13], [12], [47], namely, as electronic analogs of biological synapses connecting neuron cells. The instructors should then discuss the known functions of biological synapses and make a parallel with the properties of memristive systems.…”