Question: Has anyone consider using spiking neural networks to provide educated guesses for reasoning?
Like this thing http://amygdala.sourceforge.net/ or whatever those people who made that are doing now.
What I'm suggesting is to insert something in the backward chaining that uses pattern recognition to select a subset of rules sequence that could achieve the goal.
Here are the people listed on the sourceforge project (maybe one of them could help):
Hartmut Prochaska hapro Developer hapro at users.sourceforge.net
Kit Mitzel kmitzel Developer kmitzel at users.sourceforge.net
Matt Grover mpgrover Project Manager mpgrover at users.sourceforge.net
Rudiger Koch rakoch Project Manager rakoch at users.sourceforge.net