dbruck@ciss.Dayton.NCR.COM (Don Bruck@ciss.Dayton.NCR.COM) (08/02/89)
Please excuse my rudimentary knowledge of AI. I have just begun investigating
this area and would like to use this group to help me clear up areas that I
do not understand.
I am currently reading _Introduction to Artificial Intelligence_ by Philip C.
Jackson Jr. He refers to Turing machines that read tape. This seems to be a
bit dated. Is tape meant only as an easily understood metaphor?
My inclination is that, if I were to build a Turing machine (in software), I
would use linked lists, as a substitute for the tape in the book, to maintain
the information . Since linked lists can be interwoven there would be no
problem with n-dimensional matrices from a conceptual viewpoint.
Declare
dimension1 as DYNAMIC
value as CHARACTER
dim1 as POINTER to dimension1 (* Next member of list *)
dim2 as POINTER to dimension2 (* Adjacent member of dimension2 *)
dim3 as POINTER to dimension3 (* Adjacent member of dimension3 *)
.
.
.
dimN as POINTER to dimensionN (* Adjacent member of dimensionN *)
(I realize that this is not an accepted syntax and am only trying to convey
my ideas.)
I would like your thoughts on using different languages for development,
i.e. c++, LISP, PROLOG, etc. I am basicly interested in rule-based systems,
their incorporation into a corporate MIS environment, and the use of rules
to control a database server. I have access to 80286 and 80386 DOS machines.
Due to family considerations I cannot begin formal study at any of the nearby
universities at this time, although I plan to solidify my knowledge through
formal means when the opportunity returns. I would therefore appreciate some
opinions as to the incorporation of problem solving into my readings. Should
I get the strong theoretical base or should I reinforce the theory with
problem solving programs at an early time in my study. Do you feel that it is
important that I get into writing AI programs to deepen my understanding?
How soon? I remember that the CS people I went to school with did not start
writing programs until their sophomore years, IS people began writing in
Pascal about two quarters sooner.
I would also appreciate any recommendations of readings that can help me
through the introductory stage.
I would appreciate anyone's thoughts on the Turing questions and apologize for
going over AI101 material. The reading/development question is aimed mainly at
educators but I would appreciate the experiences and feelings of anyone.
Thanks in advance.
--------------------------------------------------------------------------
Don Bruck Don Bruck
NCR Corp. 4184 Bellemeade Dr.
Corporate Data Planning and Administration Bellbrook OH 45305
PCD 6 (513) 848-4420
1700 S. Patterson Blvd.
Dayton OH 45479
(513) 445-7603
My opinions and interests are my own.ntm1169@dsacg1.UUCP (Mott Given) (08/03/89)
From article <862@ciss.Dayton.NCR.COM>, by dbruck@ciss.Dayton.NCR.COM (Don Bruck@ciss.Dayton.NCR.COM): > ... The reading/development question is aimed mainly at > educators but I would appreciate the experiences and feelings of anyone. > I believe you learn AI much better by writing actual programs than by just reading about it. "AI Expert", a monthly magazine, has actual code accompanying many of its articles which can be downloaded from bulletin boards mentioned inside the magazine (on the masthead). Some vendors offer free demo disks which allow you to develop toy rule-based systems of up to 25 or 50 rules. Some AI texts, such as "Expert Systems: Principles and Programming," have accompanying floppy disks that have AI code you can load and run on your PC. -- Mott Given @ Defense Logistics Agency Systems Automation Center, DSAC-TMP, Bldg. 27-1, P.O. Box 1605, Columbus, OH 43216-5002 INTERNET: mgiven@dsacg1.dla.mil UUCP: ntm1169@dsacg1.uucp Phone: 614-238-9431 AUTOVON: 850-9431 FAX: 614-238-3214 I speak for myself
ari@kolmogorov.physics.uiuc.edu (08/05/89)
Turing machines are minimal machines. It was axiomatized in the simplest way to determine the properties of computation. Turing machines are asymptotically equivalent to any known serial computer. You wouldn't want to program or use such a machine, but its simple properties make it possible to prove certain problems "effectively uncomputable". If a problem is effectively uncomputable for a Turing machine, it is also true for most of our modern computers as they currently are. Some of these ideas have extended to understanding what is provable and unprovable, and equivalenced to Godel incompleteness Theorem. There are problems or decision problems which cannot be effectively computed. This is ultimately connected with the idea that consistent axiomatic systems are not complete (Cannot prove all true statments). A good book in this field seems to be "Computability and Unsolvability" by Martin Davis. (A Dover Text). It has been a while since I read this material, and I am not sure how much of it is considered "current" in AI, but the theorems and ideas in the text are timeless. ari Aritomo Shinozaki ari@kolmogorov.physics.uiuc.edu Loomis Laboratory of Physics (217)-244-1744 University of Illinois, Urbana-Champaign Urbana IL, 61801