Artificial Intelligence

Table of contents:

1. News
2. About this course
3. Schedule and reading
4. Assignments, exams, grades

---

1. News

• The exam has been corrected. Here are the answers and some comments. Exam results and assignments have been reported 2011-11-08. It will take a few days before the results are in Uppdok.
  
• An Excel-file shows how information gain is computed (for ID3).
• Assignment 3 is integrated in the tool.
  
• Slides for Learning inspired by nature added.
  
• Slides for lecture 2 and 3 added.
  
• 29/8 Assignment 2 is in place.
  
• 16/8. This page created, lectures (except guest lectures) in place. Assignment 1 is in place.
  

---

2. About this course

The Swedish name of this course is Artificiell intelligens, code 1DL340.
The course counts for 5 ECTS credits. The course plan is here.
The course page in Studentportalen.

Teaching staff

	Name	E-mail	Room	Phone
Lecturer	Roland Bol	Roland.Bol@it.uu.se	1356	018-471 7606
Assistent	Maria Andreina Francisco	Mariaandreina.Francisco_Rodriguez.3450@student.uu.se For booking a meeting, go here	1410

Guest lecturers

Michael Ashcroft
Olle Gällmo

Course start

• You cannot apply for admission (antagning) at the first lecture: apply through www.antagning.se as soon as possible (special rules apply for exchange students and first year Masters students).
  
• Students who have been admitted to the course and exchange students will be registered at the first lecture. It is also possible to register through Studentportalen. 
  
• The course will be given in English. Assignments must be handed in in English.

---

3. Schedule and reading

• Reading refers to "Artificial Intelligence" by George Luger, 6th edition.
• In 2012, this book will be replaced by Russell, Stuart Jonathan; Norvig, Peter Artificial intelligence : a modern approach 3. ed.
• "lektion" is the Swedish word for an interactive problem-solving session, where no new material is presented.

A rough time-budget, per week:

• lectures, 4h
• reading, 4h
• assignments, 4h
  

week	day	time	room	topic	who	reading	assignments
35	Tue 30	13	1211	What is AI? History of AI.	RB, OG	1, 16
	Thu 1	13	1211	Search introduction  Slides	RB	3.0-3.2	1. Short paper
	Fri 2	8	1211	Heuristic search methods	RB	4.0-4.3, 4.5
36	Tue 6	13	1211	Search in playing games	RB	4.4 + more
	Wed 7	15	1211	lektion: search and alpha-beta	RB		2. Game program
37	Tue 13	13	1211	Knowledge Representation, Planning	RB	7  (parts) 8.4
	Fri 16	10	1211	introduction to learning midcourse evaluation* introduction to probabilistic methods	RB (MA)	10.1, 10.4 5.0, 5.2, 5.3
38	Tue 20	13	1211	Bayesian Networks	MA	5.4, 9.3, ...
	Fri 23	10	1211	Learning, inspired by nature (Pdf 1.4M)	OG	11,12 (parts)
39	Tue 27	13	1211	Learning (induction, learning as search)	RB	10.2.2, 10.3, 10.6.2
	Thu 29	8	1211	Lektion: learning Information gain computation	RB		3. Bayesian Networks
40	Tue 4	8	1211	Expert systems	RB	6.2 (part) 8.0-8.3
	Fri 7	10	1211	AI in computer games	OG
41	Tue 11	13	1211	Understanding natural language Hidden Markov models	RB	15.0-15.2.1, 15.3.2, 15.4.1, 9.3.5, 13.1.1, 13.1.3
	Fri 14	10	1211	questions, solutions, course evaluation	RB
42	Fri 21	14-17		exam Gimogatan 4, room 1
Jan	Fri 13	14-17		re-exam Polacksbacken
Aug				re-exam

* The midcourse evaluation is an evaluation of how you like the course so far. It's not a test.

4. Assignments, exams, grades

Exam 091020 with solutions.
Exam 101019 with solutions.
Since the course in this form is quite new, there are few "real" previous exams.
We provide some sample exams, made up from real exam questions.
Edited version of the exam 090601. Solutions.
Edited version of the exam 080529 with solutions.
Edited version of the exam 070604. Solutions.

There are 3 obligatory assignments, as listed in the schedule. Detailed instructions for each assignment will follow.
Assignments will be graded pass/completion/fail only. The grade for the course will be determined by the exam grade.
Passing the exam gives 3 study points (=ECTS credits), the assignments 2 study points.

Assignment 1

Deadline: Wednesday, September 7, at 15.  This assignment is made individually.
Hand in through Studentportalen. If you have no access yet, hand in by email to Maria.
(The file area in Studentportalen is open until after the deadline, in case a student has a valid reason to request an extension. But remember, every delivery is time-stamped.)

Assignment 2

Deadline: Wednesday, September 28, at 17.
Hand in through Studentportalen. You are allowed and encouraged to work in pairs.
Only one copy per pair please, but make it clear who both partners are!

Assignment 3

You have received by email a key to download the Bayesian Network tool.
The assignment consists of the questions that you get when you follow the instructions to explore the tool.

Deadline: Wednesday, October 19, at 17.
Hand in through Studentportalen. You are allowed and encouraged to work in pairs.
Only one copy per pair please, but make it clear who both partners are!

Notes

• After installation, ignore the text "ALERT: This application has been unable to establish your identity. ..."
  You submit the assignment in the Student portal, not through the tool.
• At the very end of the instructions, there is a list of what the variables could mean (kind of). It clarifies that Variable 15 is True if the client turned out to be credit-worthy, False if not.
  
• Don't choose "Ad hoc" instead of "Test" - it crashes the program (at least on my machine).
• Importance sampling (tool) is the same as likelihood samling (lecture).