Conclusion
- We can bind via synaptic weight change.
- Our experiment shows a solution to the compositional
concept problem.
- It functions by rapidly forming CAs based around
the primitives to be bound and a binding area.
- The binding area uses spontaneous activation to
seed the binding, then later slowly erase it.
- The system shows the way forward to the solution of other
variable binding problems.
- This will be the basis for more complex computations using
neural-CA nets.
