Learning with Spiking Nets
I'm interested in spiking neural networks, and have a mechanism for
learning to categorise with them. There's
a page with
code and three papers .
I use NEST and PyNN to get reasonably accurate biological neural models.
There is code in the above link.
Some example ideas are
- Use the any of the three mechanisms on a novel data set.
- Explore the competition mechanism from the third paper. Here
there is a question of competition between the categorisation neurons,
via, for example, inhibition.
- Categorise using multiple systems and compare.
- Introduce a vision front end to the real MNIST task.
