Conclusion
- People think using spiking nets.
- Balanced spiking nets can be used to learn categories, though
it's unsupervised.
- I'm looking for some project students to work on this, and there are
some easy things to do like sorting out the spike source mechanism to
radically speed the simulations, running on SpiNNaker and trying to
distribute a unit across several neurons.
- I'm looking more broadly for collaboration with the spiking nets, so
do contact me if you're interested.
- Currently, the balanced spiking mechanism is not a particularly good
learning algorithm, but it's plausible that it can be extended to do
some really good things.
- Balanced spiking nets could be integrated into neurocognitive models.
