This review in Nature Reviews Physics, led by PhD student Shafin Bin Hamid, was a culmination of Dr. Sam Liu‘s PhD thesis on understanding both the challenges and benefits of non-idealities in artificial synapses for crossbar array-based neuromorphic computing. We define the ideal model of the artificial synapse, go through deviations from that ideal seen in the various nonvolatile memories and mitigation techniques, and then question what actually is an ideal synapse, since for numerous applications, deviations from this model are beneficial. We hope this will be a useful review for the community.
Read it here!
If you cannot access it from the main website, here is a read-only link as well.
