In spiking neural networks information are encoded in the form of spike times. Therefore, how we can apply supervised learning paradigm efficiently and what will be the procedure for output spike decoding?
Most of the related papers used rate coding for the output spike decoding (of course in hidden layers). There are several related research about supervised learning for SNNs. Common approaches are approximate non-differential neurons behavior to differential function and integrate temporal domain with spatial domain. I recommend you to read related papers (Hybrid Macro/Micro Level Backpropagation for Training Deep Spiking Neural Networks , Spatio-Temporal Backpropagation for Training High-Performance Spiking Neural Networks, Training Deep Spiking Neural Networks Using Backpropagation).
In attachments the most new papers (2018,2019) belong to SSN in supervised manner.
- Badges
- Science topic