Ongoing Work
- Cell Assemblies Mark 3 (Milner CA Mark 2)
- I think the real problem is figuring out reasonable models of cell
assemblies.
- So, I'm working with CAs that do something (2-3 associative memory),
but persist for a reasonable time.
- I'm stuck on making them work with STD, but they persist
for good times (3 seconds). Reactivation doesn't work well.
- I'm going to write that up, but then I'm back to find a better version,
hopefully supporting reactivation with STP and better persistence.
- We're also doing stuff with spiking nets and machine learning using
STDP.
- We'll get back to agents and better language processing once we
sort out the CAs.
- I think agents that can learn new objects and semantics in a virtual
environment are the way forward.
- We've also got a neural cognitive architecture (we can make production
rules in neurons).
