Fig. 1
From: 27th Annual Computational Neuroscience Meeting (CNS*2018): Part One
Classification, inference and dreaming of handwritten digits: A Following the transfer of the model from software to hardware, the iterative training recovered a classification rate close to the reference performance level of the ideal software implementation. B Examples of digits generated by the spiking network on the hardware. C The network was able to complete and correctly classify partially occluded digits