Catalogue of Artificial Intelligence Techniques


Jump to: Top | Entry | References | Comments

View Maths as: Images | MathML

Postulating Intrinsic Properties

Keywords: terms

Categories: Learning

Author(s): Pat Langley

Given two symbolic independent variables X and Y, and one numeric dependent variable Z, one cannot find a numeric law relating these terms (since only one is numeric). However, suppose that by holding X constant and varying Y, one observes different values for Z. Then one can postulate an intrinsic property I whose values are set to those of Z, and one can associate these numeric values with the symbolic values of Y. Since one now has two numeric terms Z and I, one can find the law Z/I = 1.0 (which is true tautologically). However, when a new value of X is examined and the same values of Y are used, one can retrieve the intrinsic values associated with Y. In this new context the ratio Z/I may take on some other constant value, which is an empirically meaningful law. This method can be used to infer intrinsic properties such as mass, specific heat, and the index of refraction from combinations of symbolic and numeric observations.



Add Comment

No comments.