CPSC 502 :: Artificial Intelligence I

Instructor: Alan Mackworth
- contact: mackat-signcs.ubc.ca
Teaching Assistant: Pooyan Fazli
- contact: pooyanfat-signcs.ubc.ca
Lectures: Mondays & Wednesdays, 9:30-11:00, MCLD 410
First Class: September 10 (Monday)
Office hours: TBA
We have a WebCT course for grades, discussion, and copies of chapters of the draft textbook.

Overview

This course is designed to be a "breadth" introduction to artificial intelligence. It will make a broad coverage of modern AI. It is suitable for those with no background AI, or with one undergraduate course in AI. It is designed for computer science students, but is also suitable for cognitive science students or those with some familiarity with algorithms, complexity, logic, probability....

Topics Covered

AI and agents: what is AI, history, dimensions of complexity
Agent Architectures: modularity, hierarchical control
Search: uninformed and heuristic search
Constraint Satisfaction: consistency algorithms and stochastic methods
Logical Reasoning: logical consequence, inference
Knowledge Representation: objects and relations, ontologies
Reasoning under Uncertainty: abduction, probability, independence, Bayesian networks
Planning: classical and decision-theoretic planning (utility theory, influence diagrams and MDPs)
Learning: classification, learning probabilistic models, reinforcement learning

Organization

For the first 9 or 10 weeks we will cover an overview of the material in lectures and assignments.
There will be a midterm on the material covered in class.
The last few weeks will cover current research topics presented by students.
Students will write a review paper based on the lecture they gave, with peer review of the papers.

Grading

Assignments

Text

We will use selected chapters from a new text that we are writing, which will be made available on a secure web site. An earlier text by Poole, Mackworth and Goebel, Computational Intelligence (OUP, 1998) covers some of the material. We will also use sections from Russell and Norvig, Artificial Intelligence: A Modern Approach, 2nd edn (Prentice-Hall, 2003) as the basis for certain background material.

Schedule

Sept 10 - AI and Agents, Dimensions, Applications, Control.
Sept 12 - Hierarchical Control, Search, Uniformed Search, Heuristic Search, Refinements, play with the CISpace Search Applet.
Sept 17 - Search (cont).
Sept 19 - CSPs and Consistency-based Methods, and see the CISpace CSP Applet.
Sept 24 - Variable Elimination, Stochastic Local Search; see the CIspace SLS Applet.
Sept 26 - Stochastic Local Search, Population-based methods
Oct 1 - Propositions and Inference: Semantics, Bottom-Up Proof, Top-Down Proof, Negation as Failure.
Oct 3 - Objects and Relations, Semantics, Logical Variables, First-Order Inference.
Oct 8 - Thanksgiving (no class)
Oct 10 - Semantic Networks, Ontologies, Knowledge Representation Issues.
Oct 15 - Assumptions, Probability, Belief Networks, Independence.
Oct 17 - Exact Inference, Approximate Inference, Probability and Time, Localization Applet.
Oct 22 - Planning: Representations, Forward Planning, Regression. Planning as a CSP. CIspace Representation.
Oct 24 - Utility, Decision Networks.
Oct 29 - Decision-theoretic Planning, Value Iteration Applet
Oct 31 - Reinforcement Learning, Q-learning Applet, SARSA-lambda applet, Multi-Agent Systems
Nov 5 - Casuality, CILog addition example,
Nov 7 - Review

CPSC 502: Artificial Intelligence I

Fall 2007