Josh,
A pen-pal - an AI/robotics guy - has been waxing enthusiastic about your
book. For him:
"the basic idea in his book is to devise what is essentially the "basic
computational unit - BCU" [this is my term, btw] that can be extended
indefinitely horizontally [in modules], and vertically [in hierarchical
levels] to larger and larger systems, in order to be able to handle general
AI problems. The problem is to get past the roadblock of scalability that
all previous AI systems have hit.
He calls his BCU thingie a SIGMA = sigma servo, which is an IAM
[interpolating associative memory] and a controller. You can spawn these
things as needed to handle larger problems. SIGMAs higher in the hierarchy
will deal with higher-level abstractions by taking outputs from SIGMAs lower
down. You can see the influence of object oriented programming here, and
also parallels to real brain organization.
He also mentions how the SIGMA would perform the "analogical quadrature"
operation of Hofstadter's Copycat system, which I'm not familiar with. I'm
not sure how well this scheme would work, but thought I'd mention it to you"
If it's not too much trouble - which it may be - perhaps you could expand a
little on SIGMA's, with an example or two, their importance as you see it,
and any links for further reading? Many thanks - and comments from anyone
else also gratefully received.
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=8660244&id_secret=71829167-e4dda4
