CS 980: Combinatorial Search and Heuristic Optimization
Professor Wheeler Ruml
How does a robot decide what to do? How does UPS route its trucks?
How do telcos configure their networks? Algorithms for problems like
these are some of the most fun in computer science and lie at the core
of artificial intelligence. By the end of this graduate seminar, not
only will you have read some of the current literature in this area,
you will have pushed forward the frontier of knowledge.
Official description: General techniques for solving shortest-path,
constraint satisfaction, and combinatorial optimization problems, and
their application in areas such as planning, robotics, and
bioinformatics. Students read the scientific literature and prepare
an original research contribution. Prereq: introductory AI or
permission of the instructor.
My current plan is that we'll all pursue research on the topic of
combining heuristic search and machine learning.
In Fall 2008, meets Mondays and Wednesdays 12:40-2:00pm in Kingsbury
N233 (note new time and room!).
Here's the general information hand-out
from the first day of class.
Schedule
Wed Dec 17, 3:30pm: Final papers due in hardcopy (2
copies, please) at my office. IJCAI formatting templates are
available here.
Mon Dec 15, 9:00-11:30am: Paper presentations (18 min
talk + 6 min Q&A).
Wed Dec 10: Discussion of paper drafts (20 min each). Reviews of
drafts due at 9am and handed back in class.
Mon Dec 8: WR presents on whatever you want (how to give a
talk?). Review versions
of final papers due (bring 3 hardcopies to class).
Wed Dec
3: Multi-TAC,
Constraints, 1996
Mon Dec 1: AMS (you get to reconstruct it yourself
from part1
and part2),
OR Letters, 1994.
[ Wed Nov 26 is on a Friday schedule. Happy Thanksgiving! ]
Mon Nov
24: LAO*,
AIJ 2001.
Wed Nov
19: LDFS,
ICAPS-06. Also, more project updates.
Mon Nov
17: Testing
Heuristics, J of Heuristics, 1996. Also, project updates.
Wed Nov
12: Lagoudakis, SAT-2001
Mon Nov
10: IDA*
prediction, AAAI-08
Wed Nov 5: ILDS, AAAI 1996
Mon Nov 3: hierarchical A*, SARA 2005
Wed Oct
29: UCT,
ECML, 2006
Mon Oct
27: Bayesian
Q-learning, AAAI '98
Wed Oct 22: pessimistic
heuristics, ECAI
Mon Oct 20: DTA*
Wed Oct 15: Pearl and Kim, Studies in
Semi-admissible Heuristics.
Mon Oct 13: Learning
from Multiple Heuristics from AAAI-08.
Wed Oct
8: Individual
presentations on project topics (problem statement, previous work,
thoughts so far).
Mon Oct
6: Parallel
Structured Duplicate Detection from AAAI-07. Note: we will meet
in Kingsbury W208 today.
Wed Oct
1: Parallel
heuristic search, 1990.
Mon Sep
29: Petrik
and Zilbertstein from ICAPS-08. Written project proposal due.
Wed Sept 24: Humphrey,
Bramanti-Gregor, and Davis from AI-IA-95.
Mon Sept 22: Chenoweth and Davis, IJCAI-91
(available in the library: Q335.I57, p198--203. Don't use the poor
copy available on-line.). You should have selected your project by
now - we'll discuss it in our individual meetings.
Wed Sept 17: No class meeting. You should have completed the `warm up
exercise' by now (any search algorithm and a selected test domain in
OCaml). Please email me a tarball of your code before Monday.
Mon Sept
15: Fast
and loose, STAIR 2008. Jordan's talk slides
are here.
In class, please say what project you're interested in so that you
can form teams if necessary.
Wed Sept 10: ARA*,
NIPS 2004
Mon Sept 8: Anytime
Heuristic Search, JAIR 2007
Fri Sept 5: You should have installed and played with OCaml by now.
See the AI
group's OCaml
notes. Feel free to update the wiki.
Wed Sept 3: You should already have
the Russell and Norvig
textbook (2nd edition). You might also want to
review How to Read a Paper
and The Task of the Referee.
Other resources