-----

CS 151 (Artificial Intelligence)
Syllabus

-----

Catalog Description

Knowledge representation, including rule-based systems and neural networks, learning paradigms, and philosophical challenges to artificial intelligence. Discussion of areas of current research: natural language processing, robotics, vision, cognitive modeling, case-based-reasoning. Prerequisite: Computer Science 80 (131 recommended). 3 credit hours.

Prerequisites

This course has a prerequisite of CS 80, which presupposes that you have also taken CS 60 and Math 55. In particular, I will assume that

• You have written mathematical proofs (perhaps not very complicated ones).
• You have at least a couple terms of programming experience and have some knowledge of basic data structures (e.g. linked lists).
• You have done some programming (perhaps not much) in a functional language.
• You are familiar with elementary logic.

Texts and Readings

The texts for this course are:

• Stuart Russell and Peter Norvig (1995) Artificial Intelligence: A Modern Approach, Prentice Hall, Upper Saddle River, NJ. (available from Huntley)
• Richard Kelsey, William Clinger, and Jonathan Rees, eds., "Revised(5) Report on the Algorithmic Language Scheme" (distributed in class).

The libraries have a variety of other books on AI, conference proceedings, etc. Some with particularly broad coverage (other textbooks, NCAI proceedings) will be on reserve in Sprague Library. Sprague also has the journal Artificial Intelligence, which publishes papers from all areas of AI.

Assignments, grading, etc

Graded work will consist of

• 4-6 assignments (45%)
• a paper summarizing several related published papers (15%)
• an implementation project, including project proposal and report(20%)
• a take-home final (18%)
• a class summary (2%)

If there are large differences in assignment length/difficulty, the longer/harder ones will be weighted more heavily.

In general, this average will determine your letter grade. However, I reserve the right to make minor adjustments to reflect special circumstances such as: dramatic change in quality of work as the term progresses, unusually poor class attendance, unusually wonderful class participation, problems beyond your control.

You are expected to attend class regularly. Occasional absences aren't a problem, but being gone a lot will affect your grade.

The class summary consists of standing up at the start of one class and summarizing briefly (e.g. 5 minutes) what happened in the previous class. This isn't graded: you get the 2% as long as you make a good faith effort.

Submitting graded work

Written work must be legible, written in literate English, and properly formatted. Code submitted for assignments must be properly commented and documented, meet basic style guidelines, and follow Scheme conventions. I have compiled a page of general-purpose style guidelines. Over the course of the term, I will give you feedback on Scheme-specific issues.

For a coding assignment, you must submit

• The code itself, in electronic form, using the submit system.
• A hardcopy summary of what the code does. This should include:
  • What the code is supposed to do.
  • How the code works (e.g. what data structures it uses).
  • Known bugs, limitations, extra features.
  • Sample runs of the code.
  • Quantitative performance data (when appropriate).

In addition, for the final project, you will need to:

• Submit a very brief statement of what you intend to do (proposal #1).
• Submit a longer and perhaps revised statement of what you intend to do (proposal #2).
• Demo your project for me.
• Include in your hardcopy summary a discussion of references (e.g. journal papers) used to design your project.

The paper, project report, and documentation for coding assignments must be typed (except that figures may be hand-drawn).

Lateness, excuses, collaboration

Assignments are due at the announced due date. If assignments are turned in late, I will take off points. I reserve the right not to accept assignments that are very late (e.g. over a week).

If you have a legitimate excuse (e.g. sickness, death in family) for turning in work late, you must speak to me promptly to negotiate appropriate arrangements. If it's a serious problem (e.g. affecting several classes, causing you to lose multiple days of work), you must also speak to the Dean of Students. Notice that I can't accept late work at the end of term without an official excuse via the Dean of Students.

You are encouraged to discuss general approaches and ideas with other students (in the class or outside), and get general-purpose help with programming (e.g. Scheme, Unix). However, you should not discuss details of your solutions, exchange literal copies of code for assignments, copy material from written sources (on-line or physical) without attribution, or the like. Such practices will be treated as violations of the honor code (or the relevant student regulations for off-campus students).

If you receive substantial help from someone, please document this in your submission. This might result in a lower grade. However, except in unusual circumstances (e.g. deliberate literal copying of code), this will prevent me from considering the help an honor code issue.

Under some circumstances, you may have legitimate reason to copy material, e.g. quoting from a published paper in a literature review, using a pre-supplied package of code to help implement your project. In such cases, you should explicitly acknowledge what material was copied and where it came from. This allows me to tell which parts of the work are your own and avoids an honor code issue.

Topics

Basic techniques

• search
• predicate logic
• automated inference
• natural language parsing
• planning

Representing objects and events

• object and event models
• object recognition
• robot task planning
• recognizing and interpreting situations
• advanced planning

Real I/O

• visual, speech, and scientific input
• robot control

Probabilistic techniques

• probabilistic reasoning
• statistic parsing and recognition

Learning

• neural networks
• genetic algorithms
• analogical reasoning

---

This page is maintained by Margaret Fleck.