Visa version
Visa
< föregående
|
nästa >
Jämför
< föregående
|
nästa >
Course PM
Introduction
This course gives a broad overview of the problems and methods studied in the field of artificial intelligence.
Please respect the code of honour.
Intended learning outcomes
After completing this course the student should be able to
-
recall and apply basic concepts in artificial intelligence
-
solve problems from the AI domain with limited resources in the form of time and computations
-
formulate and address an AI related scientific problem
-
demonstrate an insight into the risks of AI and its role in society
-
present work in writing and orally
so that the student can
-
make use of methods from artificial intelligence in the analysis, design and implementation of computer programs in academic as well as industrial applications
-
in an appropriate way present results and solutions.
Crew
- Lecturer and course responsible
- Teaching assistants (TA)
- Johannes A. Stork
- Akshaya Thippur
- Kaiyu Hang
Examination and grading
The examination consists of completing quizzes (individual), homework assignments (in paris) and a projects (in groups of four).
To receive a passing grade on the course a student needs to pass all quizzes, the home work assignment (HW1, HW2, HW3) as well as the requirements for the project. The project has to be completed in groups.
NOTE: All assignments MUST be completed in English
For all examinations, we use the CSC code of honour.
Assessment tasks
The examination for the course will consist of the following assessment tasks.
Quizzes |
We will have quizzes after key lectures to test that the students can apply the basic concepts. For some content in the course this will be the only assessment task and for others it will be a way to stimulate continuous learning and help motivate students to learn a concept before we go deeper in the course, for example, to be prepared for an exercise or assignment. A quiz has to be completed within a certain, short, amount of time and will be automatically corrected. It only require simple calculations and multiple-choice answers. |
HW1 |
Implement a game-playing agent. This requires students to solve a problem involving a search algorithm and design a heuristic function. Students are allowed to work in pairs but need to present the results individually. |
HW2 |
Implement an HMM-based agent for duck hunt. This requires the students to solve a problem involving uncertain information using HMMs. Students are allowed to work in pairs but need to present the results individually. |
HW3 |
Write an essay on ethics connected to the risks of AI and its role in society. |
Project |
The project must be performed in groups of four students. The students form groups themselves. The students will be given five example tasks with grading criteria specified for each tasks. The tasks will still be open enough for the students to have to formulate their specific problem but it will be clear what is required and how we should assess the students. They have to formulate a scientific problem and implement a prototype. They need to write a report and present the work orally. They have to ask questions after the presentation and give feedback on the presentation and assess it. For higher grades a student also needs to give feedback and assess another group’s report and work. |
Course components
The course has two course components
-
INL1 (Quizzes, HW1, HW2, HW3)
-
PRO1 (Project)
They are both reported into Ladok as P/F. Some additional information about the course components is maintained in CSC’s result reporting system rapp (https://rapp.csc.kth.se/rapp/).
Criteria based grading
We make use of a criteria based grading system. You will not collect points as in most other courses. Instead, to reach a certain grade you should show that you have fulfilled the criteria for that grade. In the tables below for lectures the means of examination has been indicated.
The matrix below shows which assessment tasks, assesses which ILO.
ILO |
Quizzes |
HW1 |
HW2 |
HW3 |
Project |
|
1 |
recall and apply basic concepts in artificial intelligence |
X |
||||
2 |
solve problems from the AI domain with limited resources in the form of time and computations |
X |
X |
X |
||
3 |
formulate and address an AI related scientific problem |
X |
||||
4 |
demonstrate an insight into the risks of AI and its role in society |
X |
||||
5 |
present work in writing and orally |
X |
X |
X |
X |
The matrix below shows to which level a certain ILO is assessed. The requirements for ILOs that are only assessed at an E level need to be met to pass the course but they do not influence the final grade in any other way.
ILO / Grade |
E |
D |
C |
B |
A |
|
1 |
recall and apply basic concepts in artificial intelligence |
X |
||||
2 |
solve problems from the AI domain with limited resources |
X |
X |
X |
X |
X |
3 |
formulate and address an AI related scientific problem |
X |
X |
X |
X |
X |
4 |
demonstrate an insight into the risks of AI and its role in society |
X |
||||
5 |
present work in writing and orally |
X |
ILO 1: recall and apply basic concepts in artificial intelligence
Grade |
Criteria |
E |
Can answer questions regarding all parts of the course material. |
This translates into passing all quizzes.
ILO 2: solve problems from the AI domain with limited resources in the form of time and computations
Grade | Criteria |
E |
Solves problems involving combinatorial search and probabilistic reasoning at the most basic level. |
C |
Solves problems involving combinatorial search and probabilistic reasoning applying some more advanced techniques and easily outperforming basic level solutions. At least one of the HW1 and HW2 passed before the deadline, the project passed on time and all but two quizzes passed before the deadline. |
A |
Solves problems involving combinatorial search applying some more advanced techniques and easily outperforming basic level solutions. Solves problems using probabilistic reasoning combing several methods to make good use of the information at hand. The student can show how the development was guided by comparing different designs / solutions and can assess and recommend different designs for different modifications of the problem. All tasks except one quiz passed before the deadline. |
A student that falls short of some but not all requirements for A but meets all requirements for C can be given a B for ILO2 and similar when being between E and C and D.
The requirements regarding timely completion of tasks captures the fact that solving a problem under time constraints is harder. To get a B you need to meet the time criteria at the A level. To get a D you need to meet the time criteria at the C level.
It is the state of HW1 and HW2 before the deadline that counts in the grading when considering the time constraint. That is HW1 and HW2 handed in before the deadline and assessed to be at an E level cannot be raised above E level after this. Handing in HW1 before the deadline and assessed to be at level C can still give an overall C even if HW2 is handed in later assuming it is assessed at least at level C. Level A and B can only be reached if the homework assignments are handed in before the deadline and are at that level.
ILO 3: formulate and address an AI related scientific problem
These grading criteria will be specific per project domain. See the project page for more information.
ILO 4: demonstrate an insight into the risks of AI and its role in society
Grade | Criteria |
E |
Can independently identify risks of AI and discuss these and the role of AI in the society. |
Assessed in the ethics essay handing (HW3).
ILO 5: present work in writing and orally
Grade | Criteria |
E |
Can explain own solutions orally and can write short reports discussing the results of own work. Can present work of a group in writing and orally. Can write an essay in a language that is grammatically correct and understandable. |
This is tested when presenting HW1 and HW2, the project and the ethics essay.
The final grade
To pass the course the student needs to meet the requirements at an E level for all goals. The final grade on the course is given by the average grade on HW1 and HW2, rounded down but no higher than the grade interval given by the project. The grade interval is further explained at the project page.
For example getting A on both HW1 and HW2 and C-A on the project would give an overall A, whereas getting A on HW1 and B on HW2 and C-A on the project would mean an overall B.
Self study lectures
The student body of the course has drastically different background knowledge. Since the course started we spent time introducing basic concepts in search and probabilistic reasoning. Starting this year, to be able to use the time we have in the classroom to introduce concepts which are new to must rather than only some we will provide some material for self studies. The initial two self study lectures covers two things that all CS students at KTH will already know about and are not directly related to the course material but which you need to know about.
Content | Chapters | Downloads |
---|---|---|
Code of honour (QUIZ) | N/A | |
How to use the code submission system Kattis | N/A | |
Basic search (QUIZ) | 3 | |
Basic probabilistic reasoning (QUIZ) | 13 |
Lectures
The table below gives a rough overview of the lectures. The lecture notes will be posted in conjunction with the lecture. The exact content of each lecture is only tentative and may be adapted along the way. The means of examination of the content is listed for the lectures.
The chapters refers to the course book "Artificial Intelligence: A Modern Approach" (3rd Edition) by Stuart J. Russell and Peter Norvig.
Exercises
Exercises/tutorials are carried out in smaller groups and allow for more interaction between students and teachers. The exercises also help bridge the gap between what is discussed in the lectures and the assignments but adding some more practical advice.
# / Week | Tutorial Title | Content | Downloads | |
---|---|---|---|---|
1 / w36 | Search |
Search |
||
2 / w37 | Games | Games: minimax, alpha-beta, heuristics, testing |
||
3 / w39 | HMM1 | HMM: core concepts and algorithms: forward pass, backward pass, baum-welch; solving examples |
||
4 / w40 | HMM2 | HMM: Viterbi algorithm, HMM examples, Real world HMM problems, HMM problem decomposition |
||
5 | TBD | TBD |
Additional material
The material below is not directly related to AI but we believe that the content is something that students of computer science should know about and which will help you in your work.
Help Sessions
We offer help sessions which usually in room 22:an (304), at Teknikringen 14. The KTH webpage has a map that can show you the way there. On some occasions the help sessions have to be moved to a different room. To know when we change the room, read the information page about help sessions.
Homework
This course gives a broad overview of the field of AI. The homeworks are intended to give the student a chance to work with the material a bit more hands-on.
There will be 2 homework assignments focused on implementation (HW1+HW2) and an essay focused on ethics (HW3). Please pay attention to the dates for handing them in as we will be strict with the deadlines!
Project
The project offers you a chance to practice team work and work on a slightly larger problems that require you to work together. Being able to work in a group is an instrumental skill in most jobs both in academia and industry. You will be required to write a report, make a presentation and review another group's report and act as opponents at their presentation.
The project work should be carried out in groups of 4 people. You should form your own groups. People that have not formed groups by the deadline will be assigned groups.
NOTE: You will be given a deadline to form groups. If you want to be in a certain group, make sure to form the group before the deadline. If you are not a member of a team (enough if team members have invited you to be considered to be part of the team) after that deadline you have forfeited your ability to chose group and will be placed in a group and we will assign a group for you and you will have to live with that.
Consultation
If you have questions or problems of organizational or formal character regarding this course, please do not contact the professor directly. Instead make use of the consultation time on Monday, 12:15 pm - 1 pm, Computer Vision and Active Perception Lab, Teknikringen 14, Plan 7, Room 721. Take the stairs to Plan 7 and ring the doorbell or call someone to send the elevator down.