CS 1573 Syllabus

Syllabus for CS1573
AI Application and Development

Last edited 2/14/05

Pre-requisites

CS1571: Mastery of Russell & Norvig's AIMA textbook, chapters 3, 4.1, 4.2, 7, 8 and 9
Python: Mastery of basic Python programming (see http://www.python.org to download Python, tutorials, IDEs, manuals, etc.)

Pedagogical objectives

Ability to build simple versions of a few AI applications in Python
Familiarity with full-scale versions of the same applications and their underlying AI technology
The AI application areas focused on are:
- Knowledge representation
- Natural language processing
- Planning

Facts

Professor: Kurt VanLehn, LRDC 740, 412-624-7458, vanlehn at cs-dot-pitt-dot-edu, Office hours: by email appointment, or call first, then drop in.
TA: Brian Smyth, SS 6503, 412-624-7253, bps at cs-dot-pitt-dot-edu, Office hours: Monday & Wednesday, 11 am to 2 pm
Text: Artificial Intelligence: A Modern Approach. Russell and Norvig.
Class location: The first floor conference room of LRDC.
Class time: Mondays and Wednesdays, 3-4:15 pm

Assessment policies

Your course grade will be based on class participation (25%), three programming projects (15% each), and 3 written exams (10% each).

Group work: The programming projects are designed to be feasible for a single person to accomplish. However, they can easily be extended, which makes them much more interesting. If you would like to work in a group of 2 or 3 people, then please email me within 2 days of receiving the assignment a proposal for how you will extend the project to make it suitably challenging and interesting for your group. I will probably offer you a counterproposal, and we will negotiate a suitable project.

Late policy: Programming projects must be submitted at the beginning of class; late projects will not be graded and will receive a zero. If you cannot turn attend class that day (e.g., you are sick, visiting graduate schools, attending a funeral), then submit your solution before class electronically. If you are unable to attend the exam due to medical, familial or other emergencies, please contact the instructor on or before the day of the exam in order to arrange a private exam session.

Oral exams: In order to determine how well you understand the programming project that you have submitted, the TA and the instructor will conduct oral exams shortly after the projects are submitted. During the oral exam, we will ask you questions about how the code works, why you chose the design you did, what alternatives you considered, etc. Your score on the programming project will depend in part on your answers. Oral exams will be conducted during the question period of class in a room near the classroom. If you did your project as part of a group, the whole group will be given an oral exam at the same time, and each member will be asked to described what part of the project they did. If you did your project individually, you will be given an oral exam individually.

Plagarism policy: You should do your own programming for the programming projects. Except as noted below, copying code written by others, including code you've found on the web, is considered plagarism. Hiring others to do your programming is considered plagarism. If I decide that you have plagarized your programming assignment, I will give you a zero on it. If you plagarize more than once, then I will consider formal disciplinary action, including failure of the course. Exceptions: You definitely should include code from the textbooks' Python code repository. As you will see, many of the functions you will want to use have been stubbed out--you will have to provide their implementation, but key data structures have been defined for you. If you wish, you may include code from the Python package index or the Vaults of Parnassus. As of 1/3/04, there does not appear to be anything useful in either the Python package index or the Vaults of Parnassus. However, if you find something useful (modules are added daily), you may use it if

You clearly identify in your source code where the package came from
You are prepared to answer questions during the oral exam about how it works
You contact me as soon as you have found a package that you want to use. Send me email with a URL for the package. If I believe that it simplifies the project, then I'll send email to all the students and may extend the project to make it sufficiently challenging. If your project is submitted with a reference to such a package and you did not contact me, then your score on the project will be lowered.

Class participation: During class, I will ask questions about the material you have been reading. For each question, I will randomly select a student to answer it. If you are present and make a strong effort to answer the questions, displaying knowledge from your readings, then you will receive a "1" for that question. If you are absent, do not try to answer the question or seem not to have read the assigned material, then you will get a "0" for that question. By the end of the semester, you will probably have been asked about 50 questions. Your class participation score is the proportion of 1s.

Schedule

The "pre-class" column describes tasks you should complete before class. When numbers appear, they refer to a range of pages in the textbook that you should read and master before class. TBD means "to be determined." The first section of the course (5-Jan to 7-Feb) covers knowledge representation and some specialized search algorithms. The second section (9-Feb to 16-Mar) covers natural language processing. The third section (21-Mar to 20-Apr) covers planning. The final exam (also known as Exam 3) will be scheduled later and will cover only the section of the course on planning.

Date	Pre-class	Class	Slides
5-Jan		Review first order logic	S1
10-Jan	272-310	Review theorem proving	S2
12-Jan	320-333	Ontologies, situation calculus	S3
19-Jan	334-344	Event calculus, beliefs	S4
24-Jan	344-354	Semantic nets, description logics	S5
26-Jan	110-119	Local search algorithms	S6
31-Jan	137-151	Constraint satisfaction problems	S7
2-Feb	Study	Exam 1 review	S8
7-Feb	Study	Exam 1
9-Feb	790-806 Proj. 1 due	Grammars, parsing	S9
14-Feb	806-818	Definite clause grammars	S10
16-Feb	818-823	Disambiguation, Discourse	S11
21-Feb	834-844	Language modeling, Info retrieval	S12
23-Feb	845-850 Handout	Text classification, Info extraction	S13
28-Feb	Handout	Dialogue management	S14
2-Mar	Handout	Atlas, a real dialogue system	S15
14-Mar	Proj. 2 due	Exam 2 review	S18
16-Mar	Study	Exam 2
21-Mar	375-387	Planning operators, state-space search	S19
23-Mar	387-395	Partial order planning	S20
28-Mar	395-409	Planning graphs, propositional planning	S21
30-Mar		Cancelled
4-Apr	417-430	Scheduling, Hierarchical planning	S22
6-Apr	430-440	Hierarchical planning, Conditional planning	S23
11-Apr	441-449	Execution monitoring, Continuous planning	S24
13-Apr	449-454	Multi-agent planning, Educational applications	S25
18-Apr	TBD	Intelligent tutoring systems	S26
20-Apr	Proj. 3 due	Exam 3 review	S27

Projects

The three programming projects are briefly described below. When all three are completed, you should have build a small fragment of Terry Winograd's famous SHRDLU system. All three projects use the same blocks world task domain.

Knowledge representation and reasoning Given a query expressed as a literal or conjunction of literals, return a list of all possible answers, where each answer is expressed as bindings for the variables in the query. Use a Horn clause representation of knowledge. Complete assignment is here.
Natural language understanding. Given a string such as "Is the big red block on top of the green block?", translate it into a first-order logic query and submit it to the question answerer developed during project 1. Use a definite clause grammar. Complete assignement is here.
Planning. Given a command such as "Put the small green block on a white block," plan a sequence of primitive actions for a simulated one-armed robot, execute the actions and display the new state of the blocks world. Use a partial order planner. Complete assignment is here.