Subject Objectives
A student completing 6.034 will be able to:
|
SES #
|
|
|
|
OBJECTIVE |
|
|
|
|
1 |
|
|
|
Explain the basic knowledge representation, problem solving, and learning methods of Artificial Intelligence |
|
|
|
|
|
|
|
2 |
|
|
|
Assess the applicability, strengths, and weaknesses of the basic knowledge representation, problem solving, and learning methods in solving particular particular engineering problems |
|
|
|
|
|
|
|
3 |
|
|
|
Develop intelligent systems by assembling solutions to concrete computational problems |
|
|
|
|
|
|
|
4 |
|
|
|
Understand the role of knowledge representation, problem solving, and learning in intelligent-system engineering |
|
|
|
|
|
|
|
5 |
|
|
|
Appreciate the role of problem solving, vision, and language in understanding human intelligence from a computational perspective |
|
And many 6.034 students will, as measured by exit survey:
|
6 |
|
|
|
Develop an interest in the field sufficient to take more advanced subjects
|
|
Desired Subject Outcomes
A student completing 6.034 will be able to:
|
SES # |
|
|
|
OUTCOME |
|
|
|
HOW MEASURED |
|
|
RELATED OBJECTIVE |
|
|
|
|
|
|
|
|
1 |
|
|
|
Predict the behavior of forward-chaining and backward-chaining rule-based systems |
|
|
|
H, Q |
|
|
1, 2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
|
|
Predict the behavior and estimate the cost in time and space of various heuristic and optimal search methods (depth-first, breadth-first, hill-climbing, branch-and-bound, and A*), and choose the appropriate method for particular problems |
|
|
|
H, Q |
|
|
1, 2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
|
|
|
Predict the behavior of various constraint-satisfaction methods (backtracking, forward-checking, constraint propagation), and choose the appropriate method for particular problems |
|
|
|
H, Q |
|
|
1, 2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
|
|
|
Develop small rule-based and search-based expert systems, predict performance characteristics, and describe the role of rule-chaining and search in intelligent-system engineering |
|
|
|
P1 |
|
|
3, 4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
|
|
|
Use rules and frames to represent behavioral, classification, and causal knowledge |
|
|
|
H, Q |
|
|
1, 2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
|
|
|
Apply basic machine learning methods such as nearest neighbors, identification trees, neural nets, and genetic algorithms |
|
|
|
H, Q |
|
|
1, 2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
|
|
|
Predict the behavior of the basic machine-learning methods, and choose the appropriate method for particular problems |
|
|
|
H, Q |
|
|
1, 2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
|
|
|
Modify and extend simple implementations of the subject's representations and methods |
|
|
|
H, Q |
|
|
3, 4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
|
|
Develop small learning systems, predict performance characteristics, and describe the role of learning in intelligent-system engineering |
|
|
|
P2 |
|
|
3, 4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
|
|
|
Discuss key issues in knowledge representation, problem solving, and learning |
|
|
|
P1, P2 |
|
|
1, 2, 3, 4, 5, 6 |
|
Key:
|
|
|
|
H |
|
|
|
Homework |
|
|
|
|
|
|
|
Q |
|
|
|
Quizes, Final |
|
|
|
|
|
|
|
P1 |
|
|
|
Project 1 |
|
|
|
|
|
|
|
P2 |
|
|
|
Project 2 |
|
How do the versions of 6.034 differ?
6.034, fall version, versus 6.034, spring version:
- Professor Patrick H. Winston is in charge in the fall, Professor Tomas Lozano-Perez in the spring.
- Grade distributions do not differ significantly.
- The spring version will be considerably different from a year ago, as it was taught experimentally for the first time in 2002, and Professor Lozano-Perez contemplates substantial changes.
- A large fraction of the material taught covers the same ground. The most conspicuous differences are that the fall version focuses toward the end of the semester on models of aspects of human intelligence and the spring version includes a major section on formal logic.
- The spring version places somewhat greater emphasis on programming assignments.
- The fall version includes two one-hour lectures, one one-hour tutorial, and one one-hour recitation; the spring version offers two one-and-a-half hour lectures and one one-hour tutorial.
6.034, 2002 version, versus 6.034, 2001 version. :
- In contrast to the 2001 fall version, students in 2002 will not be required to submit weekly lesson plans. In their place, we expect to require a few pages of analysis on assigned readings from time to time.
- As in 2001, a special "intensified" section, in which students do projects in addition to the other work in the subject will be offered. Students in the projects section will earn three units of extra credit.
What will the subject cover this semester?
Will there be a textbook?
No. In our survey of year 2000 students, the textbook was rated dead last in value, behind the on-line tutor, live tutorials, recitations, and lectures. A substantial fraction of the material in the subject is discussed in Artificial Intelligence, 3rd edition (Patrick H. Winston, available at Quantum Books, Amazon.com, ...), which may be helpful to those students who like to learn from textbooks, but the cost is outrageous, and most year 2000 and 2001 students did not touch it.
How will the subject be structured?
During a typical week, there will be:
- Two lectures
- One human tutorial
- One on-line tutorial
- One problem set
- One recitation
Am I expected to attend lectures, tutorials, and recitations?
Yes. We believe that the lectures, tutorials, and recitations are an important part of the MIT experience, and we work hard to make them interesting and useful.
- Lectures introduce powerful ideas and relate the material to the "big picture." We often include questions on the quizes and final that you can answer only by faithful lecture attendance.
- Tutorials provide you with an opportunity to ask questions and to demonstrate your understanding. A substantial part of your grade is determined by your tutorial participation.
- Recitations play a major role in clarifying the material and demonstrating how problems are solved.
When are the quizes?
For the quizes, see the
Lecture Schedule.
How are final grades to be computed?
We expect to form a weighted sum as follows:
- 0.3 x your numeric score on final
- 0.25 x your numeric score on quiz 1
- 0.25 x your numeric score on quiz 2
- 0.2 x the score assessed for your tutorial participation and off-line homework (if any)
Then, we will combine the result with your performance on the on-line problems, which will be represented by a score from zero to one. We expect you to work the on-line problems until you get them right, thus earning a score of one.
Combine means multiply by (min~1.0~(sqrt~(/~s~0.9))). This function looks like this:
![curve2](../../../../NR/rdonlyres/Electrical-Engineering-and-Computer-Science/6-034Artificial-IntelligenceFall2002/9B0961BB-F175-45D9-8642-99E05E69B849/0/curve2.jpg)
Thus, you will want to do all the on-line problems faithfully, both because of the direct impact those problems will have on your grade and because of their close connection to material that will appear on examinations.
What is the final grade distribution likely to be?
Following institute guidelines, we give grades based on our assessment of how weighted sums translate to qualitative performance, rather than a curve (see the back of one of your previous grade reports).
Results from previous years indicate that the distribution of grades will approximate the following distribution, but note that the actual boundaries may be higher or lower.
20% A
30% B
40% C
We discuss the bottom 10% and borderline cases at length at the end of the semester.
Do I need to know how to program in Scheme?
The subject is not centered on programming, but a substantial fraction of the homework requires an understanding of Scheme and working out some small Scheme programs. The quizzes and final do not include questions centered on Scheme programming, but they sometimes include questions described, in part, in Scheme. Veterans of 6.001 should have no trouble with the programming involved; students skilled in some other programming language will have to spend a couple of weekends reading the Scheme book and should work out a catch-up plan with their tutor; students with no programming experience are advised not to take the subject.