Course COM316 Artificial Intelligence:
MW 11:50 - 1:05 Dilley Room / Unix Lab
Gary Parker
Winthrop Annex (ext. 5208)
parker@conncoll.edu
Text:The Scheme Programming Language, Third Edition by R. Kent Dybvig
Software:Our primary Scheme programming environment will be Petite Chez Scheme using SWL which will be available on all Unix/Linux workstations and can be obtained from Art Potter. Petite / SWL can be downloaded for use on PCs from: http://www.scheme.com, but this version may not run the program we use in this course
Web Page:http://cs.conncoll.edu/com316
Winthrop Annex: Monday, Wednesday: 1:30-3:30
This course uses the incremental learning of solutions for progressively harder problems to cover a breadth of concepts used by researchers in their attempt to develop an artificial mind. General areas covered include search techniques, propositional and first order logic, representation, production systems, planning, learning, and connectionist systems (neural networks).
Artificial Intelligence (AI) deals with a vast array of subjects that are often seemingly unrelated. Going from formal logic systems, to vision processing, to planning, to robotics, to artificial neural networks; it covers several well developed areas of study. The central idea of all these is the emulation of a living creature. Not necessarily a model of one that exists on earth, but one that has the characteristics of a living creature; perception, movement, learning, reasoning and interaction. One or all of these can be incorporated in a study of AI. Although all these areas are important for the formation of an artificial creature, in this course we will concentrate on the areas that are directly involved in the creation of an artificial mind. This has been the central theme of AI, hence the name. We will try to cover the breadth of this area to give the student a good overview of research done.
Drawing an analogy between "ontogeny recapitulates phylogeny" and what we call the "AI skills development recapitulates the history of mind design" we will cover areas of AI dealing with mind design by tending to follow its history. This should take us step by step through the evolution of concepts appropriate to create an actual artificial mind. We will learn progressively more complete systems, always striving for the final yet still unreached goal of an artificial human brain. The way the student should approach this course is to consider what is needed to make each step and what is the key element that limits each step.
The idea of this course is for us to reason through intermediate steps on the way to an artificial mind. We will somewhat follow the history of AI to help guide us and keep us in unison as we progress toward an artificial mind. The accuracy of our portrayal of history will not be as important as its use in our learning of the concepts of AI, but some idea of history's general trend will be a natural artifact of our learning.
The typical Monday class will start with discussions of solutions to the last programming assignment, the reading, and the three questions (listed below and on the main web page). AI Notebooks will be due on this day. We will then introduce the next step in our progression to an artificial mind. A problem will be introduced that is beyond our current capabilities to handle. Questions will be answered to ensure all understand the problem. During the Wednesday class meeting student solutions will be handed in and discussed. Students are to work individually on this coming up with their solutions. No outside sources should be used, only the tools and background knowledge provided by the instructor. This HW should take approximately 2-3 hours of thinking and 15-20 minutes to type the answer. Do not do your thinking at a desk; do it while you play soccer, eat, or go to a party. It is more important to have a good answer that you can defend, than a correct answer.
The Wednesday class will start with a discussion of student solutions. Each student should be prepared to defend what they came up with. We will discuss the standard solution and reading material will be distributed. The next phase will be the implementation of the standard solution; although students who can convince the instructor that their solution is as good as or better than the standard will be allowed to implement it. Students will be provided with program segments written in scheme that will provide the base for solving the problem and make the general problem concrete. Questions will be answered to ensure all understand the use of support programs.
For each problem give your solution and answer these three questions:
1. What other types of problems can we solve using this method? In other words, the problem probably deals with a very specific situation. Can we categorize what general category of problems this method can solve?
2. Assuming that this solution does not give us a fully autonomous artificial mind, what is holding us back?
3. Can we restate this problem and/or add more tools to gain more ground in our search for the artificial mind? What small change will force us to develop a solution that is one step closer to a fully autonomous artificial mind?
Problem Solutions |
25% |
Answers to Questions |
10% |
Class Participation / Presentation |
30% |
Programming Assignments |
35% |
AI Notebook |
+/- 10% |
Problem Solutions:Normally due each Wednesday. The first is due on 19 Sep. These can be done individually or in groups of no more than two. Using outside resources is not allowed.
Answers to Questions:Normally due each Monday The first is due on 24 Sep. These are to be done using no outside resources. You should think about these and then discuss them in groups of no more than four. These are NOT to be handed in individually, unless you disagree with your group.
Class Participation / Reading Discussion:Each student will lead a discussion of the assigned reading. All students are required to do the reading and will be prepared to discuss it. A large part of the learning experience will come from discussion of the problems. The participation grade will be determined by the quality of your participation in the problem and reading discussions.
Programming Assignments:There will be AI programming assignments normally due every week. Scheme programming assignments will be due at every class for 2 or 3 weeks. You need to do the Scheme programming assignments individually. You may work in teams of no more than two on the AI programming assignments.
AI Notebook:This notebook is for you to organize and summarize what you have learned about AI. It is to have a summary paragraph about each topic that we cover, plus a copy of the readings marked with your notes and/or comments. These are to be done individually.
Notes:
If you have a physical or mental disability, either hidden or visible, which may require classroom, test-taking, or other reasonable accommodations, please see me as soon as possible. If you have not already done so, please be sure to register with the Office of Student Disability Services, at Extension 5428 or Extension 5240.
The Roth Writing Center provides one-to-one peer tutoring (free of charge) to help student writers of all abilities during all stages of the writing process. To make an appointment, call x2173 or stop by the Writing Center at 214 Blaustein. If you're a confident, experienced writer we can help you to push your ideas and polish your style; if you're a relatively inexperienced and not-so-confident writer we can also help you, by working on grammar or organization or whatever you need. Writing Center tutors are trained to help you to discover what you think through writing. Working with a tutor gives you the opportunity to share your work-in-progress with an actual reader, so that you can get useful feedback on that work before you have to turn it in for a final grade. For further information, visit the Writing Center web page at http://write.conncoll.edu/.