Neuro-inspired Speech Recognition with Recurrent Spiking Neurons

摘要: This paper investigates the potential of recurrent spiking neurons for classification problems. It presents a hybrid approach based on paradigm Reservoir Computing. The practical applications are limited due to their non-trivial learning algorithms. In Computing, instead training whole network only output layer (known as readout neurons) trained. These neural networks termed microcircuits which viewed basic computational units in cortical computation. connected columns linked with other neighboring areas. read out information from each and can serve both reservoir readout. design space this is split into three domains; front end, reservoir, back end. work contributes identification suitable end processing techniques along stable compact dynamics, provides reliable framework related