Generative Systems
- Each of these systems are complex.
- From a finite number of tokens, they can generate an infinite
number of structures.
- That is they're generative.
- So, they can be represented by grammar rules that we're familiar
with (e.g. context free rules).
- Jackendoff notes the variable binding problem for neural systems,
but I think we neuro-computational researchers have answers to
this.
- My parser is an example of a neural generative system.
- That is, we can implement these systems.
