Neuromorphic electronic circuits for building autonomous cognitive systems

摘要: Several analog and digital brain-inspired electronic systems have been recently proposed as dedicated solutions for fast simulations of spiking neural networks. While these architectures are useful exploring the computational properties large-scale models nervous system, challenge building low-power compact physical artifacts that can behave intelligently in real-world exhibit cognitive abilities still remains open. In this paper we propose a set neuromorphic engineering to address challenge. particular, review circuits emulating synaptic dynamics real-time discuss role biophysically realistic temporal hardware processing architectures; challenges realizing spike-based plasticity mechanisms real present examples implement them; describe recurrent networks show how Winner-Take-All working-memory decision-making mechanisms. We validate approach with experimental results obtained from our own systems, argue presented work represent components efficiently elegantly implementing cognition.