Other Future Work
- Associative memory. Kailash Nadh, the symbolic PP attachment
guy, is a PhD student who
has just started on this.
- We had to make the parser small and fast because we are
running into limitations in the number of neurons we
can simulate. So we want to fix our capacity problem.
Some options are:
- Write it better (say in C with smart memory handling).
- Distribute it.
- Hardware; I'm hoping to take advantage of some neural
simulation chips.
- Wait for better standard processors.
- Cognitive Modelling.
- Learning language.
- Biologically realistic vision.
- Modelling neural behaviour with fLIF neurons.
- Neuro-Cognitive Architectures.
- New domains, learning new domains
- Turing test
