CSE571 Artificial Intelligence
Nov 12th Notes
William Cushing
0) Returned midterms
1) Went over questions
a) Question 5. Liberal grading, one answer in book – so not going over in class
b) Question 4. Also liberal grading, as there are many ways of answering it.
As long as it works, it works – so good grading.
There was supposed to be an abnormal predicate – for the discussed way Of
handling normative statements. So some points could be lost there.
c) Question 3. The separators in the representation were periods, not commas.
i) For the correct stratification, the answer set is the empty set.
For the incorrect version of the program (reading the program wrong) the
answer set is {r,p}.
ii) For the correct reading, the answer set is {q}
For the incorrect reading, the answer set is {r,p}
iii) For the correct reading, the answer set is {r}
For the incorrect reading, the answer set is also {r}
d) Question 2. Almost everyone got this right – typical example was (a if not b, b
if not a).
e) Question 1. Herbrand Universe is all of the objects; including those produced
by functions. The Herbrand Base is just all of the predicates applied to all of the
objects (ie: Base = Pred(Universe))
i)
HB={p(a),p(b),p(c),q(a),q(b),q(c)}
LM={q(b),q(c),p(a)}
AM={q(b),p(a),p(b),q(c)}
ii)
Add f(a),f(b),f(c),f(f(a)),f(f(b)),f(f(c)), ... to the HB above
Add q(f(a)) to LM above
f) Question 0 – take-home. Graded liberally, despite the fact that it was take-
home and people could have read the referenced sections (5.1, 5.2).
The basic argument is that the answer sets do not encode a plan because the plan
does not work in every possible initial state – there are multiple possible initial
states (uncertain information), and a plan needs to work in all of them.
About 30% of people had a good argument and example, and got 20 – others had
some thoughts, but not enough to get more than a 10. The last portion had a good
example, but didn’t abstract out well enough, and got about 15.
2) Bayes nets
Covering 2 chapters out of a different textbook:
Artificial Intelligence: a new synthesis. Nilsson. Chapter 19, 20.1.
Chapter 19 is available in the library (Noble) on hold. (by Thursday).
“see how much you remember from last class” ==> bigger example:
