Old course page. The new one is under construction

CMPUT 657 Heuristic Search with Applications to Games

Winter 2018, Instructor: Michael Buro
Lectures: MW 12:30-13:50 CSC B43, F 9:30-10:50 CSC B43

eClass Forum
Course material (assignments, lecture notes) (id/passwd required, revealed in first lecture)

News

Paper Selections

Tentative Schedule

(Lectures are front-loaded to give students the chance to apply all lecture materials to their projects)
                     Lectures                     
Week of M   M (+0)      W (+2)       F (+4)    Topics
 1. Jan 08  L1        L2/A1r         L3        Intro, Game AI Successes, MiniMax Search
 2. Jan 15  L4          L5           L6        AlphaBeta Search, Transposition Tables, Move Sorting
 3. Jan 22  L7          L8           L9        Playing with Search Windows and Search Depth
 4. Jan 29  L10         L11          L12       Evaluation Functions, Parameter Optimization
 5. Feb 05  L13/A2r     L14        L15/A1d     Single-Agent Search, A*, IDA*, Pattern Databases
 6. Feb 12  L16         L17          L18       Hierarchical/Triangulation/Multi-Agent Pathfinding
 7. Feb 19  ------- READING WEEK -------------
 8. Feb 26  L19         L20        L21/A2d     Chance and Sampling, Monte Carlo Tree Search, UCT, AlphaGo
 9. Mar 05  L22/CPA     L23          L24       Imperfect Information Games, Card Game AI, RTS Game AI
10. Mar 12  CPR          -            -
11. Mar 19   -           -           PAP
12. Mar 26  PRO          -            -
13. Apr 02   -           -            -
14. Apr 09   -          PRP1      PRP2/PRR

Legend:  Ajr / Ajd : assignment j released / due (before lecture)
         Li        : lecture i
         CPA       : choose recent research paper (Mar. 5)
         CPR       : choose project topic (Mar. 12)
         PAP       : paper presentations (Mar. 23, 9:00-10:50) and summary due
         PRO       : project progress report (Mar. 26, 10% malus if not sent in time)
         PRPi      : project presentations (Apr. 11/13)
         PRR       : project report due (Apr. 13)

Course Overview

Search is at the heart of artificial intelligence (AI) research. AI applications often have to search among the alternatives for either the optimal answer (optimizing) or the best result given limited resource constraints (satisficing). This was best epitomized by the chess match between Deep Blue and Garry Kasparov. The computer, searching 200 million chess positions per second, narrowly edged the human world champion (~2 chess positions per second).

This course will cover many important search algorithms used in AI ranging from single-agent A* search, over two-player search (alpha-beta), to Monte-Carlo Tree Search (MCTS). Algorithms will be evaluated in terms of their algorithmic complexity, implementation considerations, utility, interaction with application-dependent knowledge, etc. At the end of the course students will know how video game engines find shortest paths quickly, how strong board game, card game, and video game AI systems work, and what current research challenges in this AI area are. Course projects can become seeds for theses!

There will be 3 assignments and a project in the course. Two assignments are designed to be fun and competetive: 1) Writing a program that plays a two-player perfect information game. 2) Writing a single-agent search program to solve a puzzle. Programs will be evaluated (in part) by competing in a round-robin tournament, allowing each student's program to test its ability against all other programs. In the second half of the course, students will summarize a recent research paper of their choice and choose a project to work on (if more than 10 students take the course, projects will be done in teams of 2 if possible). Project presentations will finish the course. There is no final exam.

Tentative Course Syllabus

Suggested Reading

The course is mostly self-contained. Links to research papers will be provided.

Prerequisites

Familiarity with Java or C/C++ and fundamental algorithms.

Grading Scheme

In this course grades will not be curved, they are absolute - following these cut points:

>= 95% A+   >= 90% A    >= 85% A-
>= 80% B+   >= 75% B    >= 70% B-
>= 65% C+   >= 60% C    >= 55% C-
>= 50% D+   >= 45% D    <  45% F

Please visit this page to learn about our interpretation of letter grades. I have the discretion in setting the borderline between passing and failing and, in doing so, may consider a students entire performance across the term as well as their overall percentage.


Academic Integrity

The University of Alberta is committed to the highest standards of academic integrity and honesty. Students are expected to be familiar with these standards regarding academic honesty and to uphold the policies of the University in this respect. Students are particularly urged to familiarize themselves with the provisions of the Code of Student Behaviour and avoid any behaviour which could potentially result in suspicions of cheating, plagiarism, misrepresentation of facts and/or participation in an offence. Academic dishonesty is a serious offence and can result in suspension or expulsion from the University. (GFC 29 SEP 2003)

Copying and cheating on assignments will be penalized with a mark of 0 (see the standard handouts for academic dishonesty and copying and cheating), and Section 30.3.2 Inappropriate Academic Behaviour.

Course Policies

Unless otherwise noted, the CS Department Policies are in effect.

Collaboration

Students are encouraged to discuss and solve problem sets in small groups to speed up learning and stimulate idea exchange. In the end, however, students must write down their own solutions and be able to solve similar problems independently. You must give credit to any source that substantially assisted you in completing the assignment. A source includes fellow students, books, papers, TAs, and me. Failure to give proper credit is considered plagiarism.


last modified on  ; you are visitor # since Aug/31/2017