Uncertainty in a numeric concept discovery system
This paper deals with the problem of reasoning about conceptualizations (sets of relevant parameters) of physical processes. It discusses the problems faced by the COPER discovery system [Kokar, 1986a, 1986b]. COPER conjectures parameters characterizing physical processes and the functional relationships among them. It must handle two kinds of uncertainty — about relevance of parameters, and noise. The paper shows that the straightforward probability/statistics approach is only partially adequate for this problem and indicates some of the possible ways to solve the discussed decision problem.
Unable to display preview. Download preview PDF.
- Birkhoff, G., (1960). Hydrodynamics. A study in logic, fact and similitude., Princeton University Press, Princeton.Google Scholar
- Drobot, S. (1953). On the foundations of dimensional analysis. Studia Mathematica, 14, pp.84–89.Google Scholar
- Kokar, M., M. (1985). On invariance in dimensional analysis (Technical Report MMK-2/85). Boston, MA: Northeastern University, College of Engineering.Google Scholar
- Kokar, M., M., (1986a). Determining Arguments of Invariant Functional Descriptions, Machine Learning, 1, pp.403–422.Google Scholar
- Kokar, M., M., (1986b). Discovering functional formulas through changing representation base. Proceedings of the Fifth National Conference on Artificial Intelligence, Philadelphia, PA, pp.455–459.Google Scholar
- Lewis E.E., (1987), Introduction to Reliability Engineering, Wiley, New York, N.Y., p. 61.Google Scholar
- Mendenhall, W., Scheaffer, R., L. and Wackerly, D., D. (1986), Mathematical Statistics with Applications, PWS Publishers, Boston, MA, p. 253.Google Scholar
- Whitney, H., (1968), The Mathematics of Physical Quantities, part I and II, American Mathematical Monthly, pp. 115–138 and 227–256.Google Scholar