learning and knowledge respresentation

[tomcat]

Hi Pei,

The NARS paper discusses "adaptive or learning system must also be a
self-organised system - the system can accomodate itself to new
knowledge ..."

Are there any examples which shows the system can learn? If not, any
help to come up with a simple example would be great.

I also guess the knowledge representation is through feeding the
system with the facts which are stored in the memory. Is this right or
am I over-simplifying it?

cheers
tomcat

[Pei Wang]

In NARS, "learning" and "reasoning" (and many other capacities that
are traditionally named differently from each other) are carried out
by the same process, which produce new beliefs and modify existing
beliefs. The details are discussed in Chapter 11 of my book, as well
as in the following publications:

The logic of learning
by Pei Wang
Working Notes of the AAAI Workshop on New Research Problems for
Machine Learning, Pages 37-40, Austin, Texas, July 2000
[available on-line]

A logic of categorization
by Pei Wang and Douglas Hofstadter
Journal of Experimental & Theoretical Artificial Intelligence, Vol.18,
No.2, Pages 193-213, 2006
[not available on-line]

For the simplest examples, search for "induction" and "revision" in
the example files.

All knowledge is indeed initially feeding into the memory, but the
system is selecting and changing the pieces all the time, rather than
simply keep them as in a database.

Pei