Spiking neural networks advantages over "traditional" NNs

Hey, this is a great question. I did some research on spiking neural networks for my masters degree 4 years ago...and haven't touched them again until about a month ago when I figured out what I wanted to do with them again.

1. One nice feature of spiking neural networks is the inherent time-based behavior. If you know how to train or build the network, you should be able to get precisely timed responses to inputs. That said, I haven't seen any method for training SNNs that rises above the rest. Google Scholar and IEEE explore are good resources.

2. Nope

3. The work done by Izhikevich is the most interesting that I've run across. He's done some great research on biologically accurate and computationally efficient neurons and networks, and I don't see any sign of it stopping.
   Izhikevich publications

1- I am not sure there exist as generic learning rules for SNN as there are for MLP. SNN will certainly require much more simulation time (they are in the milliseconds time-scale)

2- I don't know of any, but see suggestion 3

3- I stronly suggest Theoretical Neuroscience by Dayan and Abbott. It covers a broad range of models from firing-rate model (ML) to spiking models. Computational Explorations in Cognitive Neuroscience (O'Reilly and Munakata) is also interesting, though very different.