CSC384S - Introduction to Artificial Intelligence

Paulo Pacheco ppacheco@cs.toronto.edu

Office Hours: TBA; by appt.
Please send email to the address above and not to my CDF account. Include CSC384 in the subject line.

Please be sure to attend your tutorials. You are responsible for material covered there. You will normally hand in assignments in tutorial (in the correct tutorial section). Our two midterm tests will be held in tutorials as well.

New Links and Info

News and other Material:

• May.07: Final Marks. Please let me know ASAP of any problems.
• Apr.21: Check out the "Final Exam & Midterm" section. Don't forget to also check this.
• Mar.27: Assignment 5. Due date is April 7.
• Mar.13: Assignment 4. Due date is March 26. Please check the assignments section for extra files.
• Mar.09: Second test confirmed: March 21, in tutorial. Asst4's due date was tentatively scheduled for March 24, so we will move it to March 26 to give you a bit of time to breath in between.
• Mar.09: Correction in asst3 to furthest predicate. Click here for more details.
• Feb.25: Assignment 3. Due date is March 10. Please check the assignments section for extra files.
• Feb.19: Link to Eric Joanis' tutorial web page.
• Feb.13: Test cases for assignment 2 are now available.
• Feb.09: Assignment 2.
• Feb.04: Handout: Manhattan Bike Courier
• Feb.04: Handout: interpretations
• Feb.04: Midterm 1 is on Feb.07 - Topics go up to LCFS (inclusive). Here is a helpful link.
• Feb.04: Some typos have been fixed - see assignments section.
• Feb.04: Assignment 1 is due tomorrow until the end of lecture - PAPER ONLY - Late assignments won't be accepted.
• Feb.04: Test cases for the assignment 1 - click Here.
• Jan.28: Some typos have been fixed - see assignments section.
• Jan.28: First test confirmed: Feb 7 (in tutorial).
• Jan.28: Newsgroup: ut.cdf.csc384h.
• Jan.28: Assignments dates have been updated (Assignment 1 due date is Feb 5 - handed in on paper only).
• Jan.22: Assignment 1.
• Dec.25: Jan 6 - First lecture.

Click here for a list of CDF accounts that have been created for 384. If you have a permanent account on CDF, then you will also be listed here. If your name is not on the list and should be, send your instructor email (or otherwise get in touch). NOTE: The list has been updated as of Dec. 25.

The course newsgroup is ut.cdf.csc384h and is accessible on all CDF and CSLab machines. Feel free to post questions there, especially those whose answers will be of general interest to the class. Watch the newsgroup for announcements as well (as well as the course Web page).

This course provides an introduction to AI, covering some of the fundamental topics that underly the many different areas of artificial intelligence, as well providing an introduction to several of the more specialized areas of AI.

The emphasis will be on the logical foundations of AI and problem solving methods. We will cover some of the widely applied techniques of logical representation languages and search techniques for problem solving. These tools are used throughout many areas of AI. In addition, we will spend some time dealing several more specialized topics: game tree search; action representations and planning; probabilistic reasoning and decision theory; Some Prolog programming will be required as part of the course. The course is not, by and large, a programming course.

The topics we will cover include:

1. Introduction to AI and subareas of AI
2. Logical Reasoning with Definite Clauses
3. Problem Solving: Graph-based Search Techniques
4. Actions and Planning
5. Probabilistic Reasoning, Bayesian Networks, and Decision Making

The course material will be covered primarily in lectures and tutorials. You should read the appropriate sections of the text prior to attending that lecture, and then review the text again, along with any additional notes you took, after the lecture. I will make my lecture slides available online in PDF format (see below). These are not intended to replace your lecture notes, but supplement them and help you organize them. To the extent possible, I will try too post lecture slides before class; however, you should not rely on this fact. They will certainly be posted within a couple of days of the lecture, after I have a chance to correct the inevitable typos that you find in class :-) Keeping up with the reading is imperative!

There will be five assignments given in the course. Most but not all of these will involve Prolog programming. The approximate out and due dates for the assignments are:

• Assignment 1: Out: Jan.20; Due: Feb.5
• Assignment 2: Out: Feb.3; Due: Feb.24
• Assignment 3: Out: Feb.24; Due: Mar.10
• Assignment 4: Out. Mar.10; Due: Mar.24
• Assignment 5: Out: Mar.24; Due Apr.07

There will be two midterm tests of 45 minutes duration. These will be held in tutorial. They are tentatively scheduled for Feb.07 and Mar.14. The official midterm dates will be announced at least two weeks in advance.

There will also be a final examination to be scheduled by the registrar.

The grading scheme for the course is as follows:

• Assignments (5): 35% (7% each)
• Tests (2): 20% (10% each)
• Final Exam: 45%

Course Material:

The textbook for CSC384 is Computational Intelligence: A logical approach, Oxford University Press, 1998, by Poole, Mackworth and Goebel. This will be the main source for the course. You will also need to rely on lecture material. If necessary, we will supplement this with secondary references and online material. There are currently some supplementary references on Prolog, which the the programming language we will use in the course. Since you've all seen Prolog prior to taking CSC384, these should be viewed as reference material only. Please keep an eye on the List of Course Materials and References. We will keep you apprised of any specific material you should read.

Lecture Slides, Readings, Online Notes

Lecture slides from all past lectures are posted here. They will sometimes be posted before class, but this is not guaranteed. Please keep an eye on this space to see what readings from the text you are reponsible for before class. These slides should not be used as a substitute for attending lectures and taking notes. And they do not take the place of reading the assigned Chapters in the text.

If I fall behind (say, by more than a few days) in getting slides posted, please make a fuss :-)

Links to AI Resources and other Fun Stuff

As we proceed through the course, we will sometimes discuss various research projects, ideas, robots, games, etc. We will add links to appropriate material on the 384 Links Page as we encounter it. Feel free to browse these at your leisure.

Assignments

Assignment 4

Late assignments won't be accepted.

• Iterative deepening regression planner: idrplan.P
• Example domain with coffee-fetching robot: miniPlanDomain.P
• Example planning problem for coffee-fetching robot: prob.P
• Test cases

Assignment 3

Late assignments won't be accepted.

• Interactive game playing shell: play.P
• Game specific predicates: othello.P
  
  

Assignment 2

Late assignments won't be accepted.

• A* code, which you should test and modify. Should run under most any version of Prolog (but use XSB on the CDF machines, that's what I tested it on recently).
• IDA* code, which you should test and modify. Should run under most any version of Prolog (but use XSB on the CDF machines, that's what I tested it on recently).
• My member and notmember predicate definitions, which can be used in multiple path cycle checking. You can use XSB library "member" predicate also from the "basics" library.
• My exp2 predicate definition to compute powers of 2 for the h3's implementation.
• Correction q8: prove that
• monotone restriction holds for h1.
• admissibility and monotone restriction do not hold for h2 and h3.  Since the monotone restriction cannot hold when admissibility doesn't hold, it is sufficient to give a counter example that infirms admissibility.
• Test cases: they are now available.
  
  

Assignment 1

Late assignments won't be accepted.

• Typo question 4d: the clause should be has_keys(X) <- prof(X) instead of has_keys(X) <- faculty(X). The typo is corrected in the online version of the assignment.
• Typo question 2b: the predicate you should define is totalCredits(L,TotCred) instead of enoughCredits(L,TotCred). The typo is corrected in the online version of the assignment.
• Typo question 1b: I forgot to tell you what \pi(higher_than, CDN$, US$) maps to. It maps to true (i.e. \pi(higher_than, CDN$, US$) = true).
• Clarification question 2: When a course is taken twice, it could be because a student failed the course, withdrew from the course or simply took the course a second time to improve his/her (already passing) grade.
• Clarification question 2: Do not bother checking the validity of the knowledge base.  You can safely assume that the transcript and program requirements are specified using predicates consistent with the syntax and definitions described in the assignment.
• Test cases question 2: they are now available.
• Correction in test cases kb2 and kb3: the handout says that you can assume that equivalent courses always appear together in a group, but that wasn't the case for kb2 and kb3.  Click here for the corrected versions of kb2 and kb3.
• Clarification question 2: you can use any built-in operator/predicate of XSB prolog.  Note however that you can easily do the assignment without using them.
• Clarification question 4: when asked to give an SLD derivation, write only the "path" of logical consequences (along with the clauses selected and the chosen substitutions) that proves the query.  You don't need to write the unsuccessful paths nor the backtracking you may have gone through before finding the proof.
• Clarification question 2: if a course is part of two groups then passing this course counts towards the requirements of both groups.
• Practice exercices: Here are some exercices (with solutions) that may be helpful to practice for the midterm or to answer some questions on Asst1.  The exercices of sessions 1 and 2 are relevant to the course. Note however that the midterm will cover other topics that are not in those exercices, such as search.

Final Exam & Midterms

• Final exam date: see schedule
• Duration: 2 hrs
• Material subject to examination:
• All assigned readings
• All assignments
• All lecture slides
• Review checklist of topics covered in the course and material for which you are responsible on the final exam.
• Bring a non-programmable calculator.
• No other aids allowed (i.e., no notes, no lecture slides, no textbook, no computer, etc.)
• To practice, check out
• the textbook's web site for exam style questions
• previous final exams available online at the library's website
• Try this.
• Good luck to everyone!

• Mar 21 - Midterm 2 - in tutorial
• Duration: 45 min
• Material subject to examination:
• Search, planning and games (note that games are not covered in the textbook so you should rely on the lecture slides).
• Slides for lectures 7-16
• Textbook Ch. 4.1-4.6 (excluding dynamic programming), 8.1-8.3 (excluding event calculus and partial order planning)
• No aids allowed (i.e., no notes, no lecture slides, no textbook, no calculator, no computer, etc.)
• The test is now marked.  Here are the solutions with the marking scheme.

• Feb. 7 - Midterm 1 - in tutorial
• Duration: 45 min
• Material subject to examination:
• Slides for lectures 1-9
• Textbook Ch. 1, 2, 3, 4.1-4.4, 4.6(cycle checking and multiple path pruning only)
• No aids allowed (i.e., no notes, no lecture slides, no textbook, no calculator, no computer, etc.)
• Here are some exercices (with solutions) that may be helpful to practice for the midterm or to answer some questions on Asst1.  The exercices of sessions 1 and 2 are relevant to the course. Note however that the midterm will cover other topics that are not in those exercices, such as search.
• Solutions for Midterm 1

Marks

• Final Marks.