More Words
- Of course this is really a poor way to encode words.
- Your brain didn't assign words by reading word net.
- People do some really cool symbol grounding.
- I think this is really the cutting edge of AI.
- When we can solve it, we'll probably have gotten to
Turing test passing AI.
- We'll talk more about this with learning
.
- Pulvermuller notes that different types of words are in different
(overlapping) brain areas.
- Most of the symbol grounding stuff is on nouns.
- Verbs are also important, and involve argument structure.
- CA wise overlapping encoding has problems with CAs recruiting each
other.
- We've made progress with a compensatory learning rule.
- Others use global inhibitory pools.
