Artificiel intelligence, ai09

The course gives a broad overview of the problems and methods studied in the field of artificial intelligence.

Please respect the code of honour.

Patric Jensfelt: Responsible and lecturer
Danica Kragic: Lecturer
Javier Romero: Course assistant

Aim| Literature| Examination| Lectures

Homeworks| Project| Help | Course evaluation

One of the biggest changes from previous years is that we will use BILDA this year for posting the lecture slides and for submitting the homework and projects.

Aim of the course

After completing this course the student will be able to

• analyse and solve problems involving various forms of search algorithms, including the design of heuristic functions to improve the efficiency of such solutions
• formulate and process knowledge in propositional and first-order logic
• formulate and solve problems with uncertain information using Bayesian approaches
• explain the basics for communication between agents and work with grammars to parse and generate languages and compute models for probabilistic language processing
• relate the material to some of the currently active research problems in computer vision and robotics
• develop systems that utilize artificial intelligence.

Besides the purely technical aspects, the course also provides the student with

• practical experience in working with problems both individually and in groups
• experience in presenting results both in writing and orally
• experience with project based work

Literature

The course literature is
Artificial Intelligence: A Modern Approach (Second Edition) by Stuart J. Russell and Peter Norvig, Prentice Hall (2003), ISBN 0-13790-395-2

Examination

The examnination consists of completing homeworks (combination of individual and pair work) and a projects (in groups). Starting from 2007 all grades at KTH are goal oriented. This means that to receive a certain grade the student has to fulfill the requirements for that grade. To recieve a passing grade the student should show basic knowledge of the course material and be able answer specific questions for which an answers are given during the lectures or in the textbook. A passing grade also require the student to work in a group, write a report that presents the results and present the results orally and critally review another groups report. For higher grades the student must show ability to formalize and apply the knowledge on practical examples.

A grade A-F is given on the course.

• Each homework assignment (INL1-3) and the project (PRO1) will be given a grade A-F
• To pass the course all homework assignments and the project need to have a passing grade (A-E).
• To receive the higest grade a student should show that all goals for all different parts of the course have been fulfilled at the level A. However, to increase the margin somewhat we use a system where the final grade is calculated as the mean grade on the results rounding to the closest grade down. That is, the partial results A B A A will result in the final grade A whereas A B B A will result in B.

For all examinations, we use the following code of honour..

Lectures

The official schedule for the lectures can be found here in text format and graphical format.

Lecture overview

The table below gives an overview of the lectures. The lecture notes will be posted in conjunction with the lecture at BILDA. The exact content of each lecture is only tentative and may be adapted along the way if something takes more or less time than planned for example.

NOTE: We will go through the homeworks in class so that you get quick feedback. We will not have had time to correct and grade them until then as the deadline typically will be 24h before that but you will see some suggested solutions. We will nto hand and out the solutions so you need to be there if you want this feedback.

NOTE: You are NOT allowed to print the notes at CSC (printing at home of course OK). They are here for your convenience and we trust you to honour your part of this by not printing them.

Since not everyone can come into BILDA we have alos put the lecture slides here

	Date	Content	Chptrs
L1	Aug 31	Introduction and agent models	1-2
L2	Sep 1	Search algorithms	3,4
L3	Sep 9	CSP and Games	5,6
L4	Sep 10	Logic, inference and know.rep.	7-10
L5	Sep 16	HW1 feedback Probabilistic reasoning	13-15
L6	Sep 17	Probabilistic reasoning	13-15
L7	Sep 23	Probabilistic reasoning Planning	13-15
L8	Sep 24	Planning HW2 feedback	11-12
L9	Sep 25	Making decisions	16,17
L10	Sep 30	Communication	22,23
L11	Oct 1	Vision and robotics	24,25
L12	Oct 9	HW3 feedback Summary	all

Homework

Goals

This course gives a broad overview of the field of AI. The homeworks are intended to give the student a change to work with the material a bit more hands-on. To get the highest grade on the homework you will have to show that you can apply the knowledge in practice by solving some practical problems.

Homework assignment

There will be 3 homework assignments handed out during the course. The table below lists the tentative dates when the homeworks will be available for download and when they are due. The homeworks should be posted in BILDA (http://bilda.kth.se). The deadlines may change (only forward in time). The deadline that matters is the one in BILDA so pay close attention.

The homeworks consist of 3 parts (A, B and C). Part A should be solved indiviually. This means that we expect to get different formulations from everyone. You are allowed to discuss with others but the answer should be yours alone. Identical or near identical answers will lead to loss of points or worse. Part B involves an implementation and can be carried out in pairs. Part C consists of a review of someone elses Part B. The idea being that even if you did not make an implementation yourself you get to see one, if you solved it you get to see someone else take on the same problem and you get to practice the important skill of judging the quality of work of others.

The homework should be handed in in English. Homeworks handed in in other languages will not be graded. Please direct any questions about the homeworks and electronic handins to the course assistant Javier (jrgn AT csc.kth.se).

NOTE:The three parts of the homework should be handed in seprately. Make sure to post your answers in the right place in BILDA.

NOTE:Check BILDA to see how to proceed if you want to make a review (Part C) but not make an implementation (Part B).

	Out date	PartA+B due	PartC due	Downloads
HW1	1/9 after class	9/9	14/9	homework1.pdf
HW2	10/9 after class	17/9	22/9	homework2.pdf
HW3	17/9 after class	5/10	8/10	homework3.pdf

Project

Goals

During the course you will learn about many things but as the scope is so broad you will not have time to focus so much. The project offers you that chance. It also gives you a chance to work in a group, write a report and make a presentation of your work at the end. For some students this could be an important rehearsal for the thesis work were you will be graded not only on the content of the thesis but also on the report, presentation as being able to critically assess the quality of someone elses work.

SUGGESTION: When working in a project it is a good idea to (s)elect a leader that will coordinate things. This person should make sure that things happens, calls meetings, etc.

This year the project will be about implementing an agent for Sokoban.

Report

The project work has to be presented in a form of a report. Remember that this document is the main way in which your ideas for the project can be communicated in detail. The report thus largely defines what you have done. This means that you cannot expect someone to read your code line by line to find all the fancy stuff you have done. The report should consist of the following:

• a title page with your names and birth dates
• a brief summary of your project and obtained results (Abstract)
• short introduction to the problem and a description of the relationship to the course material. also present what ytou think are your biggest contributions so that this is clear. be sure to credit anyone that you got help/code from
• a clear formalization of the problem using course material: which methods, how and why
• a summary of your design approach and implementation
• a thorough analysis of your experimental results

You should remember that your report is the main method of communicating to the course instructors what you have accomplished and what methodology you have chosen to take. It is therefore very important that the report is well organized and well written.

The report should not be longer than 10 pages and in pdf format. Reports not in pdf will not be graded. The page limit is to force you to express yourself in a concise manner and for your reviewers not to have to read too much. The report should be posted in BILDA together with the source code and instructions for how to build and run it in a zip-file. See BILDA for more instructions.

Presentations

Each presentation is give a slot of 30min. This includes 15min presentation and 10-15min of questions/discussion. Remember that the presentation is part of the grade on the project so please make an effort here. A projector will be available for the presentations. if necessary a laptop will also be made available so that you can bring a USB stick if you do not have your own laptop. Please send an email the day before if you want to use such a laptop . The presentation will be in seminar room 22:an at Teknikringen 14 (same room as the help hours).

Opposition

In connection with the project presentation each group will be reviewed by another group. The opposing group should read the report and ask questions after the presentation. This opposition will be graded as well. Fill in the opposition protocol before the session and BRING 2 COPIES (filled in until the presentation part), one for the teacher and one for the other group so that they get some feedback on their work. This protocol is available in 3 different formats ( pdf , word and openoffice )

Grade on project

The grade for the project is calculated similar to the master thesis grade. The work itself, the report, the presentation and the opposition are all graded. Each of these gets grade 0 (fail) - 3 (excellent). Each part need to have at least grade 1 (pass) for the project to get a passing grade. The grade for the work itself is multiplied with a factor of 3 and the points are summed up.

• E: 6-
• D: 8-
• C: 10-
• B: 13-
• A: 15-

Groups

The project work should be carried out in groups of 4 people. You are free to form your own groups. People that have not formed groups by the deadline will be assigned groups. See BILDA for more information.

Important dates for project work

The table below lists some preliminary dates for the project work

Date	Task
September 10	Form groups
October 16	Project reports due

Help hours

If you have questions about projects, homework or the course material you can come ot the help hours

Date	Room
7/9 12-13	Seminar room 22:an, Teknikringen 14, level 3
17/9 9-10	Seminar room 22:an, Teknikringen 14, level 3
30/9 12-13	Seminar room 22:an, Teknikringen 14, level 3