15-885 ADVANCED TOPICS IN AI/COMPUTATIONAL SCIENCE:
Computational Scientific Discovery
(X-listed 80-514 - Philosophy of Science Seminar)
Instructor: Valdes-Perez (valdes@cs)
TR 1:30 - 2:50, WeH 3412
Credit: 1 Coreunit, 12 University Units
This course surveys those tasks of scientific reasoning that have been
automated in computer programs, the general methods that have been
used, and what new analogous tasks in the sciences might be attempted.
The emphasis is on high-end tasks: tasks that ordinarily are called
creative.
As introductory background, we will review (1) characteristics of
scientific research as carried out in practice, (2) some fundamental
concepts and principles of scientific inference, and (3) methods of AI
that have been most applied to scientific discovery.
Then, a series of case studies will be described from a variety of
fields, including biology (cell & molecular), chemistry, physics,
medicine, and social science. Discussions will center on (1) the
context of the scientific task, (2) the computational method that is
used, and (3) the role of prior knowledge and how it is handled.
Finally, we will cover generalizations from these case studies,
research methodology, the interface between scientist and discovery
programs, and philosophical issues.
COURSE REQUIREMENTS: If you are taking this course for credit, you
will be expected to (1) do the regular homework assignments, which
will include some hands-on experience with programs and maybe a little
programming, (2) pass a midterm and final exam, and (3) keep a notebook
and record questions that arise during your reading prior to class
sessions. These questions will be raised and discussed during class.
PREREQUISITES: Graduate student status in CS, Philosophy, or consent
of the instructor. Non-CS students should have some programming
experience and appropriate undergraduate mathematics.
TEXTBOOKS:
(1) J.E. Oliver. The Incomplete Guide to the Art of Discovery.
Columbia University Press, New York, 1991.
(2) W.C. Salmon. The Foundations of Scientific Inference.
University of Pittsburgh Press, 1966.
(3) P. Langley, H.A. Simon, G.L. Bradshaw, and J.M. Zytkow.
Scientific Discovery: Computational Explorations of the Creative
Processes. MIT Press, Cambridge, Mass., 1987.
(4) Other reading will consist of the primary literature and
shorter selections from books.
TENTATIVE OUTLINE OF COURSE:
Introduction (1 day)
Background
Textbook vs Frontier Science (2 days)
Concepts of Scientific Inference (3 days)
Heuristic Search and Representations (3 days)
Scientific Discovery Programs
Historical modelling (2 days)
Early programs (2 days)
Midterm Examination (1 day)
Recent Programs (14 days)
Generalizations, Research Methods, etc. (3 days)
Final Examination