Computational Theories of Learning and Reasoning
Fall 2001
| Course Projects |
Projects Due 12/14/01) |
General
-
In this class I do not require that you submit Term papers/Projects
proposals. However, you are welcome to give me the planned abstract
of your paper and I will comment on it.
-
The topic is of your choice, but must be related to one or several of
the papers that we covered.
-
The work can be related to your own research, or provide you an
opportunity to explore a topic that you may continue to do research
on. However, the above point is still relevant - it needs to be tightly
related to paper(s) covered in this class.
-
You may do experimental work, theoretical work, a combination of
both or a critical survey of results in some specialized topic.
- It is supposed to be a Computer Science Paper, not a Philosophy paper!
-
Try to make it interesting!
Ideas for Projects
Here are some examples. Please send me e-mail or come and talk to
me if you would like to pursue one of those or if you have an idea
that you'd like to discuss.
Each of the items on the list below is just a brief idea that can be
taken in several directions - both as a theoretical study or an
experimental study.
- Relations between model-based representations and rule
based representation. Study of knowledge representations and use in
reasoning.
-
L2R and Robust logic; relations, differences; in what ways do
they complement each other.
- Reasoning paradigms in the context of story comprehension and other NLP
problems. What's suitable and what's not; modeling and feasibility.
-
A comparative study of paradigms for probabilistic FOL representation.
-
KR languages. On the use of KR languages as feature languages. (Formal
definitions, semantics, computational properties.)
-
Robust logics vs. other FOL formalisms;
- Robust Logics: open problems and extensions.
- L2R non-monotonically: open problems and extensions.
-
Probabilistic representations: global probabilistic representation
vs. partial representations via a collection of definitions.
(e.g., the robust logic model).
-
Alternative representations of distributions. (Starting point:
Darwiche's paper;L2R paper).
Dan Roth