CSE logo

University of Washington Department of Computer Science & Engineering


 CSE 573 – Artificial Intelligence - Autumn 2003

 

 

Welcome to the 573 Home Page.

Instructor: Dan Weld

[ Administrivia | Topics & Schedule | Problem Sets | Project | Resources ]

The primary goal of this class is to provide a rigourous introduction to Artificial Intelligence, explaining the challenges inherent in building an intelligent system and describing the main techniques and tools. We will focus on search, knowledge representation, constraint satisfaction, planning with some material on learning and reasoning about uncertainty. A secondary goal of the course is to teach how to read and analyze research papers. Thus some of the reading will be in the textbook (or a survey paper) and some will be recent conference papers; in the latter case, students will write program-committee style reviews. There will be several short problem sets and a modest project.


End of quarter details

Take-home: available 2:30pm Friday 12/12 due 2:30pm Monday 12/15 Final take home exam

Final robot code: accepted immediately, but turn in no later than 9:00pm Tuesday 12/16

Report: due: 1pm Thursday 12/18 (8 page limit, 12pt font, 1” margins). Model on conference paper, but no introduction necessary (do include abstract and conclusions. Discuss the online and offline aspect of your agent; describe your use of search (if you did); Include a section on lessons learned / experiments. Include a short section explaining who did what. Points awarded for: difficulty of approach, success of approach, quality of description, evaluation of alternatives, quality of experiments, quality of discussion & lessons learned.

 


Administrivia

Class schedule: M, W 1:30-2:50 MOR 225
Office hours: Monday 3:30-4:30 or please feel free to email me for another time.

TA: Masa Kobashi, mkbsh@cs (office hours by appointment)

  • Here's the Schedule for meetings to discuss your projects.

 

The (required) textbook is Russell & Norvig AI a Modern Approach, (Prentice Hall) 2nd edition, 2003. (Note: this book is much improved over the previous edition). We will also assign readings off the WWW.

Your final grade will be assigned based on the following (tentative weighting):

  • 20% homeworks
  • 35% project
  • 25% final
  • 10% class participation
  • 10% paper summaries

Instructions for reviewing papers 


Schedule, Lecture Notes & Rough Outline of Topics

Date

Topic

Reading due

Lecture Slides

Sept 29

Introduction, administrivia, agents

None

01-intro

Oct 1

Agents; problem spaces; search (part 1)

R&N Ch 2, 3

02a-agents; 02b-search

Oct 6

Informed search; IDA*, local search; GA

R&N Ch 4

03-search II; robocode avi

Oct 8

Heuristics & adversary search

R&N Ch 6

04-heuristics;

Oct 13

Propositional logic

R&N Ch 7

05 adversary search; 05 prop logic;

Oct 15

Class canceled

R&N Ch 8; Ch 9 thru p 278; Section 10.3

 

Oct 20

Knowledge Representation (KR)

Robocode paper REVIEW

06 more logic

Oct 22

KR

“Two Theses” REVIEW

 

07 more kr

Oct 27

Learning

R&N ch 18

08 inductive learning

Oct 29

… continued

None

09 learning part 2

Nov 3

Planning

R&N Section 10.3; ch 11; R&N ch5

10 CSP + planning

Nov 5

Satplan 7 graphplan

SATplan REVIEW

11 SATplan+ Graphplan

Nov 10

Graphplan (cont), heuristics & time

 

12 planning

Nov 12

MDPs, Value & Policy Iteration

R&N ch 17

13 MDPs

Nov 17

Uncertainty planning using ADDs

SPUDD Review

14 BDDs, Bayes Nets, DBNs, MDPs

Nov 19

Abstraction & reachability in MDPs

LAO* Review

15 Abstraction & reachability

Nov 24

Partial observability (POMDPs)

R&N 17.4, 17.5 & POMDP Review

16 POMDPs

Nov 26

No class

None

None

Dec 1

Reinforcement Learning

R&N Ch 21

17 Reinforcement learning; Demo

Dec 3

Tournament 1 & Applic to Ubicomp

None

18 NL Question Answering

Dec 8

Intelligent Internet Systems

Probabilistic Cruciverbalist

19 Review & Topics

Dec 10

Tournament 2 & Review

None

None

Rough Week-by-week topic outline

  1. Overview, agents, problem spaces
  2. Search
  3. and
  4. Knowledge representation
  5. Learning
  6. Action languages, planning, compilation to SAT
  7. Graphplan and CSPs; MDPs
  8. More planning under uncertainty (ADDs, LOA*, RTDP)
  9. POMDPs, reinforcement learning
  10. Topics (applications to internet and ubicomp?)
  11. Project Presentations & Contest

 


Computer Science & Engineering Department
University of Washington
PO Box 352350

Seattle, WA 98195-2350 USA