Mind.Forth PD AI: 10.Aug.1999 Progress Report
Arthur T. Murray
uj797 at victoria.tc.ca
Tue Aug 10 07:36:25 EST 1999
http://www.scn.org/~mentifex/aisource.html Mind.Forth AI for Robots
is being updated. Here is my side of e-mail to a user of Mind.Forth:
Yesterday I used my two reference books by Dr. Ting to do some new
coding of Amiga Mind.Forth, in order to bring the Amiga and F-PC
versions of Mind.Forth into conformance with each other.
I made sure in the F-PC book that "entry" is a special word in
F-PC Forth, so I will no longer use "entry" as a flag. But I will
also stop using "alert" for the flag, because it makes it sound
as if the robot is *not* alert some of the time.
My new flag is now "attn" to show when the robot is paying
"attention" to human input via the keyboard.
The memetic beauty of using "attn" as a flag for robot "attention"
is that it will stick in people's minds, because so many letters
and memos circulate with the abbreviation, say, "attn: homeowner."
We'll get the entire U.S. programmer population saying, "Attn --
that's a flag in Mind.Forth for robots."
Now the latest news. Yesterday and today being my days off,
I have drifted back into coding Mind.Forth, after the gruesome
ordeal of spending the entire day of Mon.19.Jul.1999 doing the
massive port from Amiga Mind.Forth to F-PC Mind.Forth for IBM
I am especially keen and eager about finishing the code that
does the proper assignment of concept "weights" (activations)
so that the sentences and thoughts generated by Mind.Forth
really make sense, that is, are "true" statements.
But yesterday I started in by bringing the two Mind.Forths
into conformity, and by coding your ".uk" and ".ear" ideas.
It works! Yesterday on mon.9.aug.1999 for the first time since
tues.6.jul.1999 I went back into Amiga Mind.Forth-28 and I tried
out your suggestion, as described above.
Now it is so easy to finish a run of Mind.Forth, then type in
".uk" and see the contents of the "uk" channel array.
Question: For Forthers and non-Forthers, would it be good to offer
both options, first by coding in the complexities of
: .uk ( various display code ) ;
and by then simply making another term synonymous?
: LANGVIEW .uk ;
Or would such a redundancy slow Mind.Forth down, or use too much RAM?
For conformity, I used an idea from [a Forth expert] and I created
a variable "big" to set the size (512 instead of 999) of all
the memory-channel arrays in Mind.Forth. Now it will be easier
for any programmer to experiment with different values of the
array size by initializing the variable "big" at different values.
More information about the Neur-sci